19 July 2001
Symposium on Global Review of 2000 Round of
Population and Housing Censuses:
Mid-Decade Assessment and Future Prospects
Department of Economic and Social Affairs
United Nations Secretariat
New York, 7-10 August 2001
Statement from Japan *
Senichi Obayashi and Hideo Umezawa **
1. The population census of Japan has been conducted almost every five years since 1920. The last census was taken as of 1 October 2000; the preliminary counts of population were released in December 2000 and the results of the prompt sample tabulation in June 2001. Currently, the results of the first-stage complete-count tabulation, including the final counts of population, are being published by prefecture. In the first-stage complete-count tabulation, the basic statistics for areas smaller than municipalities are being made available combined with boundary data prepared using Geographic Information System (GIS).
2. In the 2000 census, as in past censuses, the self-enumeration method was adopted. The census employed 826,000 enumerators and 89,000 supervisors. Important features of the 2000 census are (a) efforts to elicit user needs were more extensive than in the past censuses; (b) use of optical character readers (OCRs) for data capturing; (c) extensive utilization of the imputation method based on the composition of each household; (d) diversification of media for data dissemination and (e) promoting the use of GIS in census-related work.
3. In addition to asking about data needs from such stakeholders as government organizations, local governments and the Statistics Council, a working party including demographers, economists, sociologists and statisticians was organized to discuss the census topics. Given the limitation on the number of the topics for the census, factors considered in determining the census topics were (a) necessity of investigating in the census; (b) usefulness for administrative purposes at national and local levels; (c) easy understanding of questions by all respondents and (d) public concern about privacy protection.
4. In deciding the topics and the method of enumeration, how the census is regarded by the general public is also crucial to maintaining quality of the data. For this purpose, a survey was conducted during or after the census taking to enquire of sample households about the census. Reports from the local governments were collected and claims and questions from the general public by telephone or e-mail were reviewed. These results for the 2000 census will be indispensable for planning the next census.
5. Regarding the choice among data-collection methods in Japan, the basic topics on population and households necessary to obtain the statistics for each municipality have been included in the census, and other topics have been covered by periodic sample surveys. Although the possibility of employing a long form in the census was also pursued, introduction of the sample survey method in the census was not adopted because it was anticipated that use of different types of questionnaires would be unacceptable for the residents. To utilize the administrative records of residents, the following must be carefully considered: (a) what kind of population information is needed by the users, taking into account that in many cases it is stipulated by the law that census population counts must be used, and that the population counts at the municipal level based on the administrative records are sometimes considerably discrepant with the census counts; (b) limitation of the items recorded on the administrative records, which may result in not reducing costs as expected to obtain necessary data; (c) whether it is legally allowed to use the administrative records and whether it is acceptable to the public to use them.
6. Optical Character Recognition (OCR) questionnaires were used for the first time in the census of Japan. The decision was made based on the experiences of using Optical Mark Reader (OMR) forms in the past censuses and of employing OCR-type questionnaires in other sample surveys. The use of OCR also offered the advantage of saving space on forms. OCR cannot read Japanese characters with a sufficiently high degree of accuracy at present. This is one of the reasons why automatic coding techniques for industrial and occupational classifications are not employed in the census tabulation. However, input images are displayed on monitors and used in personal computers (PCs) connected to the Local Area Network (LAN) for manual correction of erroneous data and for the coding of industrial and occupational classifications. Data capturing using OCRs was outsourced and successfully finished by July 2001. As maintaining strict confidentiality was one of the most important factors, the data-capturing work was done in the building of the Statistics Bureau/Statistics Centre.
7. Media for dissemination of the census results have been expanded along with the development of new technologies. The 2000 census results can be obtained and used not only by CD and MO but also in such ways as the Internet, the LAN system for all government departments and agencies (SISMAC), and the WAN (Wide Area Network) system for the prefectural governments (SWAN). Considering that many statistical analyses are now done using PCs and that the data format convenient for users is of the spread-sheet type, the output tables of the 2000 census are being produced mostly in EXCEL form.
8. The Census Mapping System (CMS) of Japan is built on the basis of the Basic Unit Blocks (BUBs) which were introduced in the 1990 census as permanently demarcated areas. They are used for setting up enumeration districts (EDs) and for the smallest geographic tabulation units. An important application of this system is to produce maps of the EDs by combining boundary data of the EDs with digital maps that record background information, such as topographical features. Approximately 30 per cent of the EDs were covered in the last census by this method of digital ED maps. To promote and improve this method, however, it would be essential to have maps of different kinds used for producing the ED maps prepared on the common base and to update them frequently.
9. A post-enumeration survey (PES) was conducted in the 2000 census, and it will be used in the planning of the next census as in the past. There are a number of problems in conducting the PES and understanding its results. Although it is desirable that a PES be conducted soon after the census enumeration, it is practically impossible to do so. Under such circumstances, it is preferable that the area where matching of PES results with those of the census be made is sufficiently broad, but this is also difficult under limited resources. Moreover, it is not easy to explain to respondents why the PES is necessary. After evaluating the 2000 PES results, the necessity of a PES for the 2005 census will be reviewed.
10. In the Statistics Bureau/Statistics Centre of Japan, the work to collect and review in detail the experiences in the last census is under way. Planning of the 2005 census will be accelerated beginning in October 2001, when the themes to be considered and examined for the next census will be made clear.
11. The views expressed above are those of the authors and do not necessarily represent those of the Statistics Bureau/Statistics Centre of Japan.