The first population census undertaken by the British Administration in this country was in 1891. Since then, censuses have been conducted at ten yearly intervals except in 1941 when the 1939-45 War interrupted the series. Post independence Ghana has witness three population censuses in 1960, 1970 and 1984. The 2000 census is the fourth census to be conducted since independence.
Population censuses have been conducted in Ghana at approximately ten-year intervals since 1891 except in 1941, when the series was interrupted as a result of World War II but was resumed in 1948. The first post-independence census was conducted in 1960 and the next in 1970, with the expectation that a decennial census programme would be maintained. Due to circumstances beyond the control of the statistical organization, however, the third post-independence census could not be conducted until 1984. Similarly, the next census which was expected to have been conducted in 1994 was delayed. Only in 1995 was it possible to have the needed commitment to ensure the conduct of the fourth post-independence census which was scheduled for the year 2000.
The 2000 Population and Housing Census was undertaken to update current information on the size, sex, age, composition and other characteristics of Ghana's population and to ascertain the specific changes in these characteristics which had taken place since the last census was conducted in 1984. The Census was expected to ensure the continuation of a time series of demographic and socio-economic benchmark data at the national and sub-national levels and enhance the capability-building programme of the Statistical Service.
The main objective of the 2000 Population and Housing Census was to update the statistical information on the characteristics of the population of Ghana. The 2000 Population and Housing Census was the first time a full-scale housing census was conducted with a population census in one single operation.
Kind of Data
Sample survey data [ssd]
Unit of Analysis
The 2000 Ghana Population and Housing Census covered the following topics:
- General Information
- Household Listing
- Housing Condition
- Health Facilities
- Education Facilities
Producers and sponsors
Ghana Statistical Service (GSS)
Dates of Data Collection
Data Collection Mode
Interviewing teams comprised of 5 enumerators with one supervisor per team.
For the main census, therefore, the decision was to have enumerators do the coding and have supervisor do the checking. Senior supervisors were also expected to have a sample check of the work done by supervisors.
The quality control rules for Supervisors were detailed to check the quality of work of Enumerators for the entire period of fieldwork. The control activities started with the Supervisor going round with his/her Enumerators to identify EA boundaries within his/her Supervision Area (SA) and with other SAs. In addition, Supervisors assessed the sizes and spread of the EAs and, where necessary, split them.
Supervisors' control of enumeration involved coverage, consistency and final checks, using the Field Supervisor's Record Book, in which the work of Enumerators was carefully recorded for monitoring.
With regard to coverage checks, Supervisors went round EAs and localities and counted houses, paying particular attention to corners and obscure places. Supervisors randomly selected houses in each EA, identified the households and listed all the persons who slept there on the census night. In addition, the Supervisor completely re-enumerated a randomly selected household and used the information to verify the coverage of each Enumerator's work.
Data Collection Notes
A trial census was held in 4 selected districts in July 1999 for 10 days. This was followed by the training of 5,400 trainers from January 24th- February 15th 2000. The trainers were to become supervisors of the main census.
A total of 27,100 enumerators were engaged for the census. The enumerators were sent out in teams of 5 under one supervisor. In all there were 220 senior field supervisors who worked in the 110 districts of t he country.
The fieldwork began on 19th March 2000 with identification of EA boundaries, listing of structures, enumeration of institutional population and floating population, and continued with the enumeration of the household population after Census Night (26th March 2000) for two weeks. A mopping-up exercise to cover about 10 percent uncompleted areas, mainly in Accra and Tema, was carried out until the end of April. The coverage of the population was mainly on de-facto basis, but a questionnaire item was provided to enable the usual resident population to be obtained.
Questionnaires were in the English language but enumerators, as part of their training were encouraged to use the local language when the need arose.
Consultation with Users
Work on the census questionnaire started in 1998 bearing in mind the data needs of the country. A simple questionnaire was sent to the ministries, relevant government departments, research institutions, relevant departments in the universities, private business associations and other users seeking information on the following:
· whether the organization had used any previous census data
· the specific census data used
· what use the census data were put
· any data that were needed but had not been provided in previous censuses
· general comments on population censuses.
Response to the questionnaire was encouraging; some respondents sent in the completed forms while others came over to discuss their data needs.
Selection of Topics
Selecting topics for inclusion in the questionnaire involved the review and consideration of the following:
· topics covered in the 1984 population census,
· recommended topics from the United Nations Principles and Recommendations for the 2000 round of Population and Housing Censuses,
· data requests and suggestions from users based on the answers to the questionnaire sent to them,
· list of users' requests compiled by the Statistical Service over a period of time.
A number of meetings were held at both the Census Secretariat and the Technical Advisory Committee levels to discuss the topics and requests. Decisions on topics for inclusion were based on the relevance of topics and the data needs of the country as well as practical considerations of application of concepts.
The final questionnaire consisted of 15 questions on housing characteristics and 20 questions on population covering the following areas:
· household characteristics
· geographical location and internal migration
· demographic and social characteristics
· economic characteristics
· literacy and education
· fertility and mortality.
All the population topics investigated in 1970 and 1984 censuses were maintained, because they were considered as still relevant to the country's data needs, especially in terms of maintaining a time series of socio-economic data.
The questionaires were published in English.
The Census data editing was implemented at three levels:
1. Field editing by interviewers and supervisors
2. Office editing and coding
3. Data cleaning and imputation
Data editing was partly manual and partly automatic.
Editing of the census data involved correcting errors from the field and those introduced during the capturing process. Both Structural Edits and Within Record Edits were used to clean the census data.
a) Structural Edits
- Structure edits check coverage and relationships between different units: persons, households, housing units, enumeration areas, etc. Specifically, they checked that:
· all households and collective quarters records within an enumeration area were present and were in the proper order;
· all occupied housing units have person records, but vacant units have no person records;
· households have neither duplicate person records, nor missing person records;
· enumeration areas have neither duplicate nor missing housing records.
- Each EA have the right geographic codes (region, district, locality, EA number, etc.)
- Every housing unit in an EA is entered and every record has a valid EA code
The Structural edit looked at the following situations:
· Geography edits
· Hierarchy of records
· Correspondence between housing and population records
· Editing relationships in a household
· Family nuclei
b) Within Record Edits: This consisted of validity checks and consistency edits.
· Validity checks: were performed to see if the values of individual variables are plausible or lie with a reasonable range.
· Consistency edits were performed to ensure that there is coherence between two or more variables.
The Top-down editing approach, which starts by editing top priority variables, (such as age, sex, etc.) and moves sequentially through all variables in decreasing priority was used to edit the census data.
The Hot Deck or Dynamic Imputation was also used for both missing data and inconsistent/invalid items.
The Census Secretariat carefully developed Editing and Imputation rules with written sets of consistency rules and corrections. These rules were translated into three CONCOR editing applications (Pop-Edit.exe, Hse-Edit.exe and Fertility.exe), which were used to 'clean' the data. This was done at the Regional level.
Data capture for the main census began on 21st August 2000, using three Kodak 9500D document scanners.
The data capture involved scanning of the questionnaire, interpretation of the scanned marks, transfer of the data and loading the scanned data into an oracle database. Periodic backups of the data and images were made on compact tapes.
Three 8-hour shift groups, each with a scanning assistant and a supervisor initially worked around the clock, 7 days a week for the first 4 months. Later, the duration of work was changed to a 6-day working week for the remaining 8 of the 12 months that the scanning of the census questionnaires took to complete.
A post Enumeration Survey (PES) was conducted to assess the extent of coverage and content error.