Symposium
2001/43 24 September 2001 English only |
Symposium on Global Review of 2000 Round of
Population
and Housing Censuses:
Mid-Decade
Assessment and Future Prospects
Statistics Division
Department of Economic and Social Affairs
United Nations Secretariat
New York, 7-10 August 2001
The 2000 Population
and Housing Census
in Estonia *
Mati Sundja
**
B. Census questionnaire and regulations of the 2000
population and housing census
D. Adapting new technologies to census operations
1. Compiling digital census
maps
E. Preliminary results and data dissemination
F. Some lessons learned from the census
1.
In 2000, the
first population and housing census was conducted in independent Estonia, implementing
as much as possible the United Nations Recommendations for Population and
Housing Censuses and the statistical principles of the European Union. The
reference date of the Census was 31 March 2000.
2.
In 1994, the census plan, time schedule and estimated budget for the
census were agreed upon. This formed the basis for annual budget allocations.
Several projects and developments in technical work for the census were
initiated. At the same time, the experiences and methodologies of censuses of
other countries were studied, and cooperation with organizations in other
countries was established and followed.
3.
A Government Census Committee was set up on 28 February 1995. By
resolution of the Government of the Republic, the preliminary reference date of
the census was fixed as 11 January 2000.
4.
In 1996, work on the draft Population and Housing Census Act began. The
Census Act was passed by the Parliament in May 1998. This Act established a legal basis for conducting the census.
Other relevant legislative acts were drafted and adopted in 1999 and in January
2000.
5.
In February 1999, the reference date of the Census was postponed from
11 January to 31 March 2000 at the
request of local authorities and taking into account the experiences gained
during the pilot census in March 1998. Two reasons for postponement were given.
First, elections of local authorities would have caused difficulties in organizing the census in municipalities during the last quarter
of 1999, and second, the work of enumerators is much easier in March than in
January because there is more daylight time in Estonia.
6.
The census questionnaire, regulations and forms
of census questionnaires for the 2000 population and housing census were
approved by Government of the Republic Regulation of 5 March 1999. There were
two types of questionnaires: the personal questionnaire, containing 31
questions, and the housing questionnaire with 12 questions.
7.
A user poll was carried out to find out what
census data would be needed; this helped to design the main output tabulations
and to order the software necessary for processing the census data.
8.
According to the 2000 Population and Housing Census Act, the units of
enumeration were individual persons, households, dwellings, and buildings with
one or more persons living there as of 31 March 2000. Persons were enumerated
both on a de facto basis according to their place of residence at the moment of
the census and on a de jure basis as usual residents according to their
permanent place of residence in Estonia.
9.
The population census enumeration was divided into three stages:
·
The census was preceded by a preliminary visit
by the enumerator during 26–29 March 2001. This meant that the enumerator got
acquainted with his/her enumeration area and the people living there, delivered
the leaflet introducing the census, and agreed on the time for enumeration
suitable for the residents.
·
The population census started on 31 March and
lasted through 9 April. During the census 5,400 enumerators interviewed people
all over Estonia. The enumerator visited an average of 285 persons. The census
was managed by 161 census area managers, 1,000 supervisors and 5,400
enumerators. The persons covered by the census did not fill in the
questionnaires themselves; rather, the enumerators entered the responses. This
was done to ensure the quality of the filled-in census questionnaire. Imperfect
questionnaire entries would have raised problems at the time of scanning.
·
To verify the quality of the census, a
post-enumeration sample survey was organized from 14 to 19 April; it covered 1
per cent of the total population. This was a verification survey, which meant
that in certain regions in Estonia people were visited by supervisors. During
the post-enumeration survey the supervisors asked the people to answer certain
questions (20) in the census questionnaires. This was done to verify whether
the enumerators who had enumerated the respondents had accomplished their task
properly.
10.
The Statistical Office of Estonia (SOE) launched the mapping
programme for the 2000 population and housing census in 1995. After completing
the test areas, the specifications for the digital census maps ware finalized.
According to the specification, maps at the scale of 1:50,000 in rural areas
and 1:5,000 in urban areas were drawn. The specification was optimized to
create a cartographic basis for the census planning (census area (CA)
delineation) and for the census itself (maps for enumerators, maps for
supervisors, etc.).
11. The census mapping process was outsourced from SOE. The work was done by two companies, one for urban areas and another for rural areas. The production methodology was different in urban and rural areas. In rural areas, paper maps from the 1989 census were used as a base source material, digitized by the mapping company and updated by local governments. In urban areas, the existing maps and orthophotos were used as a base source and the maps were updated by the mapping company. For rural and urban areas the municipalities compiled household lists, including the number of inhabitants in each building or apartment. The purpose of household lists was to provide information about the number of inhabitants for the delineation of enumeration areas (EAs).
12.
The
borders of census units were marked on digital population census maps, and the
maps were printed for census purposes. SOE stores digital maps of urban
areas in Mapinfo, maps of rural areas in ArcView software and household lists
in Foxpro software. The census maps were ready by December 1999. Digital population census maps with
the registered borders of administrative and settlement units are the basis for
presenting the census results in a cartographic way and for the development of
the census Geographic Information System (GIS).
13.
The
processing of the population and housing census data was completed using the
appropriate information system developed for specific purposes. This required
preparatory work as follows:
·
Designing the census questionnaires according to OCR (Optical
Character Recognition) requirements;
·
Creation of census maps;
·
Organizing extraction of data from registers in cooperation with
companies or persons maintaining the respective registers and
·
Compiling dictionaries for automatic coding (occupation, economic
activity, adminstrative unit, ethnic nationality, languages, religion,
countries).
14. Further data-processing procedures required:
·
Preparations for entering the data from census questionnaires using
Eyes & Hands software provided by a Swedish firm (ReadSoft AB) for
scanning; interpreting and correcting the letters, numbers and signs;
·
Development of an application of the software in cooperation with
AboBase Systems, Ltd., for processing census data using Oracle (automatic
coding, logical checks, etc.), application of security requirements, storage
and electronic archiving of data and creation of outputs (for disseminating on
paper, electronically and on maps); and
·
Development of a security policy for data processing and for the whole
census process in cooperation with specialists in the field. This covered
security measures ranging from the protection of enumerators to the protection
of collected personal data.
15. Data processing started in the middle of May 2000. The scanning and verification process was completed in four to five months instead of the planned seven to eight months, and the full data processing will be finished in the first half of 2001.
16. The preliminary results of the population census of total population were published by the Statistical Office of Estonia in September 2000. According to the preliminary census results, on 31 March 2000 the usual resident population in Estonia was 1,376,700. Compared with the 1989 population census, this was a decrease of 188,900 people, or 12.1 per cent. The population size has decreased due to negative natural increase (estimated to be 41,000 persons) and outmigration (85,000 persons).
17. The population of every town and rural municipality has also been affected by internal migration. Generally, internal migration is directed towards larger towns and rural municipalities around them. The departure of the former Soviet army is responsible for the steep decrease in the population size of those towns and rural municipalities where the armed forces were located.
18. The preliminary census data on population differ from the population registered in local governments, as these registers do not reflect the actual place of residence of all persons.
19. According to the preliminary estimates the undercoverage of the census is about 2 per cent. The final results will be published by SOE within two years after the population and housing census enumeration.
20. The core tabulation plan includes 176 tables. It is planned to produce all census output tables. The census results will be published and disseminated both on paper and through electronic media.
·
A census act that is too detailed and very specific may cause problems
at different stages of the census activities and may lead to
misinterpretations.
·
Changes to the questionnaire
and census procedures after the pilot census must be carefully handled and tested, and all stages of data
processing must be considered.
·
Using administrative records and databases as a source for delineation
of enumeration areas may cause uneven workloads for enumerators, especially in
the cities, due to the differences between de jure and de facto place of
residence.
·
Different software environments for storing the census data and spatial
data would be difficult to handle.
·
Producing census outputs and access of users to the results must be as
flexible as possible.
·
And, finally, a well-known problem of traditional censuses remains: it
is difficult to enumerate certain
groups of the population, such as single young mobile people, students and
others.