EAST-WEST POPULATION INSTITUTE SECOND POPULATION CENSUS TABULATION WORKSHOP

Save this PDF as:
 WORD  PNG  TXT  JPG

Size: px
Start display at page:

Download "EAST-WEST POPULATION INSTITUTE SECOND POPULATION CENSUS TABULATION WORKSHOP"

Transcription

1 EAST-WEST POPULATION INSTITUTE SECOND POPULATION CENSUS TABULATION WORKSHOP February 12-17, 1973 RECORD OF PROCEEDINGS EAST-WEST CENTER 1777 EAST-WEST ROAD HONOLULU, HAWAII 96822

2 THE EAST-WEST CENTER is a national education institution established in Hawaii by the United States Congress in Formally known as "The Center for Cultural and Technical Interchange Between East and West," the federally-funded Center is administered in cooperation with the University of Hawaii. Its mandated goal is "to promote better relations between the United States and the nations of Asia and the Pacific through cooperative study, training, and research." Each year about 2,000 men and women from the United States and some 40 countries in the Asian/Pacific area exchange ideas and cultural insights in East-West Center programs. Working and studying with a multinational Center staff on problems of mutual East-West concern, participants include students, mainly at the postgraduate level; Senior Fellows and Fellows with research expertise and/or practical experience in such fields as government, business administration or communication; mid-career professionals in nondegree study and training programs at the teaching and management levels; and authorities invited for international conferences and seminars. These participants are supported by federal scholarships and grants, supplemented in some fields by contributions from Asian/Pacific governments and private foundations. A fundamental aim of East-West programs is to foster understanding and mutual respect among people from differing cultures working together in seeking solutions to common problems. The Center draws on the resources of U.S. mainland universities, and Asian/Pacific educational and governmental institutions as well as organizations in the multicultural State of Hawaii. Center programs are conducted by the East-West Communication Institute, the East-West Culture Learning Institute, the East-West Food Institute, the East-West Population Institute, and the East-West Technology and Development Institute. Open Grants are awarded to provide scope for educational and research innovation, including emphasis on the humanities and the arts. THE EAST-WEST POPULATION INSTITUTE, established as a unit of the East-West Center in 1969 with assistance of a grant from the Agency for International Development, carries out multidisciplinary research, training, and related activities in the field of population, placing emphasis of economic, social, psychological, and environmental aspects of population problems in the Asian/Pacific area. Dr. Paul Demeny is Director.

3 SECOND POPULATION CENSUS TABULATION WORKSHOP February 12-17, 1973 RECORD OF PROCEEDINGS April 1973

4

5 CONTENTS Page LIST OF PARTICIPANTS AGENDA i i i vii RECOMMENDATIONS 1 COUNTRY REPORTS 4 Bangladesh 4 Hong Kong 6 Indonesia 8 Japan 9 Republic of Korea 9 Malaysia 10 Pakistan 10 Philippines. 11 Republic of China 12 Thailand 13 U.S. Census Tabulation Program 13 DISCUSSION 15 Computer problems in tabulation programs 15 Tabulations on residence 16 Tabulations on household, family, marriage, and relationship to head of household 17 Tabulations on fertility and mortality 18 i

6 Page Tabulations on ethnicity, race, and minority group status. 21 Tabulations on migration 22 Tabulations on education 24 Tabulations on occupation, industry, class of worker, income, and labor force participation 26 Proposal for study of labor force utilization 31 Tabulations on socioeconomic status 32 Proposed program of census data dissemination 34 Proposal for a census newsletter 35 i i

7 LIST OF PARTICIPANTS BANGLADESH f. Mr. Bahauddin Ahmad Census Commissioner Ministry of Home Affairs CANADA 2. Dr. Edmund M. Murphy Assistant Director, Regional Statistics Division Statistics Canada REPUBLIC OF CHINA 2, Mr. Tun-yih Lu Senior Statistician Bureau of Statistics Directorate-General of Budgets, Accounts and Statistics HONG KONG *f. Mr. Joseph Lee Census and Statistics Mr. Benj amin Mok Census and Statistics Department Department INDONESIA Mr. Abdulmadjid Director, Central Bureau of Statistics 7, Dr. Sam Suharto Director, Data Processing Center Central Bureau of Statistics JAPAN & Mr. Yutaka Morioka Statistical Advisor Bureau of Statistics Office of Prime Minister 1. Mr. Hideshi Honda Assistant Chief, General Affairs Section Bureau of Statistics Office of Prime Minister i i i

8 REPUBLIC OF KOREA ft Mr. Chan Wui Chief, Data Processing Division Bureau of Statistics Economic Planning Board Mr. Young Kwon Kim Chief, Census Planning Section Bureau of Statistics Economic Planning Board MALAYSIA H, Mr. R. Chander Chief Statistician and Commissioner for the Census of Population, Malaysia Department of Statistics 'It Miss Dorothy Z. Fernandez Senior Statistician and Deputy Commissioner for the Census of Population, Malaysia Department of Statistics PAKISTAN Mr. M.H.D. Hafiz Sheikh Census Commissioner of Pakistan Census Office President's Secretariat PHILIPPINES Mr. Hidalgo V. Chaves Chief Statistical Coordinator Bureau of the Census and Statistics Department of Trade and Tourism THAILAND Mrs. Ektritra Kohkongka Deputy Secretary General National Statistical Office Mrs. Anuri Wanglee Director of Population Survey National Statistical Office Division UNITED NATIONS it, Dr. Y.C. Yu Population Division iv

9 UNITED STATES IV Mr. Sam Baum International Statistical Programs Center U.S. Bureau of the Census Dr. Thomas A. Burch Chief, Research and Statistics Office Department of Health State of Hawaii M,. Mr. Robert Schmitt Department of Planning and Economic State of Hawaii Development l i. Dr. Conrad Taeuber Hyattsville, Maryland Ll t Dr. Stephen K.K. Yeh Professor of Sociology and Urban and Regional Planning University of Hawaii EAST-WEST POPULATION INSTITUTE Mr. Keith Adamson Executive Officer for Administration Dr. Fred S. Arnold Research Associate Dr. Lee-Jay Cho Assistant Director for Professional Study and Training Mrs. Minja Kim Choe Computer Programmer Dr. Paul Demeny Director Mr. Robert W. Gardner Staff Researcher Dr. Philip Hauser Senior Fellow; Professor of Sociology and Population Research Center, University- of Chicago Dr. Geoffrey McNicoll Research Associate Director, v

10 Dr. James A. Palmore Assistant Director for Institutional Cooperation Mrs. Susan Palmore Associate Program Officer Dr. Robert D. Retherford Research Associate Ms. Sandra. Ward Publications Officer vi

11 AGENDA Monday, 12 February 1973 A.M. Chairperson: Lee-Jay Cho Rapporteur: Robert Retherford 9:00 Opening ceremony and introductory remarks Paul Demeny, Director, East-West Population Institute Everett Kleinjans, Chancellor, The East-West Center Lee-Jay Cho, Assistant Director for Professional Study and Training, East-West Population Institute 10:30 Country Report of Bangladesh 11:00 Country Report of Hong Kong 11:30 Country Report of Indonesia P.M. Chairperson: B. Mok Rapporteur: Robert Gardner 2:00 Country Report of Japan 2:30 Country Report of the Republic of Korea 3:00 Country Report of Malaysia 3:30 Country Report of Pakistan Tuesday, 13 February 1973 A.M. Chairperson: J. Palmore Rapporteur: A. Wanglee 9:00 Country Report of the Philippines 9:30 Country Report of the Republic of China 10:30 Country Report of Thailand P.M. Chairperson: M. Murphy Rapporteur: Robert Gardner 2:00 The U.S. Census Tabulation Program vii

12 3:15 Discussion of Computer Problems in Tabulation Programs including programming, editing and coding, computer runs, computer capacities Wednesday, 14 February 1973 A.M. Chairperson: D. Fernandez Rapporteur: S. Baum 9:00 Discussion of tabulations on residence (de jure, de facto, and area tabulations)* 10:45 Discussion of tabulations on household, family, marriage, and relationship to head of household P.M. Chairperson: R. Chander Rapporteur Fred Arnold 2:00 Discussion of tabulations on fertility and mortality 3:15 Discussion of tabulations on ethnicity, race, and minority group status Thursday, 15 February 1973 A.M. Chairperson: S. Suharto Rapporteur: G. McNicoll 9:00 Discussion of tabulations on migration 10:45 Discussion of tabulations on education P.M. Chairperson: B. Almad Rapporteur: P. M. Hauser 2:00 Discussion of tabulations on occupation, industry, class of worker, and income 3:15 Discussion of tabulations on labor force participation Workshop sessions on Wednesday, Thursday, and Friday mornings will focus on special tabulations and analytical studies in the particular substantive areas noted for each session. Discussion will center on (1) problems and (2) issues of comparability. viii

13 Friday, 16 February 1973 A.M. Chairperson: M.H.D.H. Sheikh Rapporteur: P.M. Hauser 9:00 Presentation of proposal for study of labor force utilization 10:30 Discussion of tabulations on socioeconomic status P.M. Chairperson: S. Baum Rapporteurs: P.M. Hauser and S. Ward 2:00 Proposed program of census data dissemination 3:15 Discussion of a census newsletter Saturday, 17 February 1973 A.M. Chairperson: Lee-Jay Cho 9:00 Discussion of final workshop recommendations 11:30 Closing remarks ix

14

15 RECOMMENDATIONS Reconmiendations proposed by the subcommittee on recommendations were submitted to the Workshop-Conference for consideration and approved as follows: 1. A need exists for finding more efficient ways to tabulate small numbers of cases obtained from sample surveys. Under present programming and administrative arrangements the computer is often not so efficient for small jobs as was the older tabulating equipment. It is recommended that exploration be made by the agencies themselves, by research organizations, and by the producers of tabulation equipment of ways to expedite the handling of small tabulation tasks. It is urged that the East-West Population Institute attempt to summarize the results of such explorations and make them generally available. 2. There is an increasing need for socioeconomic-status indicators for policy and analytical purposes. Although most of the countries in Asia have not collected income statistics in their censuses, it is possible to construct socioeconomic-status indices based on census information on education and occupation, together with such proxy income items as quality of housing or possession of appliances and equipment. It is recommended that after the priority tabulations are completed, each statistical agency experiment with the construction of an index using sample data. It is recommended that a small workshop be held to: develop a useful manual on experimental procedures and to prepare a background paper for the next Census Workshop- Conference on "Post-Census Considerations 1 ' as mentioned in recommendation To improve the use of labor force information for policy and program purposes, it is recommended that each agency experiment with the tabulations required in the labor utilization approach as recommended by Dr. Hauser. 4. Because of the incomplete vital statistics and deficiencies of the demographic data in most of the countries in Asia, tabulations from the 1970 rounds of censuses of children ever born, children surviving, and own children were recommended during the first Workshop. A few countries have tabulated or presently are tabulating such data on fertility and mortality. It is recommended that each statistical agency try to produce these tabulations on a sample basis to provide basic information that will enable estimation of the level and trends 1

16 of recent fertility and mortality. In addition, each statistical agency should produce these tabulations on a sample basis by some basic socioeconomic characteristics, for the purpose of measuring differential fertility and mortality. 5. Recognizing the deficiencies of present household and family classification systems, which fail to produce a realistic picture of Asian family organization and living arrangements, it is recommended that studies be made to design a more relevant household and family typology and that, in this connection, the usefulness of the classification devised by the Organization of Demographic Associates be explored. The East-West Population Institute will provide copies of the Organization of Demographic Associates materials on classification to the countries represented at the Workshop-Conference. 6. In order to promote more comparative demographic research in Asia, more internationally comparable census tabulations are required. It is recommended that each of the statistical agencies, after completing priority tabulations, try to produce on at least a sample basis tabulations designed to effect international comparability, using the recommendations of the United Nations and such other standards as discussed in the meetings of the Census Tabulation Workshop- Conference. 7. In view of the Increasing importance of assessing the quality of census enumerations in Asia, it is recommended that each statistical agency promote research into the completeness and quality of the census enumerations and data. Exchange of the relevant information on evaluation should be promoted. 8. To assist the users and analysts of the census data it is recommended that a system of data dissemination be developed that takes into account the availability of resources in the respective countries. 9. In view of the need for a regular exchange of ideas and information on the 1970 rounds of census tabulations and publications, it is recommended that the East-West Population Institute take the initiative in developing a newsletter summarizing the relevant census activities related to the 1970 rounds of censuses, the newsletter to be published and circulated on a quarterly basis. It is recommended, in accordance with the agreement reached during the Workshop- Conference, that each country contribute a paper summarizing its census in 1970 or 1971 to a volume to be entitled, Introductions to the 1970 Censuses in Asia. 2

17 10. It is recommended that the topic of the next East-West Population Institute census conference be "Post-Census Considerations, covering the following five subtopics: (1) evaluation of the census; (2) projections; (3) measurement of fertility and mortality; (A) measurement of urbanization, and (5) socioeconomic status. It is also recommended that a background paper be prepared for each of the five subtopics and that it be distributed In advance of the conference These background papers will provide a basis for country reports. 3

18 COUNTRY REPORTS BANGLADESH Bahauddin Ahmad, country representative Robert Retherford, rapporteur Plans for a two-part Census consisting of a housing and establishment enumeration in October 1973 and a population enumeration in February 1974 are under way but not yet finalized. Although there is a long tradition of census-taking in the region that includes Bangladesh, this will be the first Census of Bangladesh as an independent nation. The government is anxious to conduct a Census as soon as possible in order to have an accurate statistical picture of the nation at its inception. In past Censuses of what is now Bangladesh, Census organizations were temporarily organized and then dismantled after the enumeration. Now for the first time a permanent Census organization is planned. The Census budget is also expected to be considerably larger than it has been in the recent past, although a tight national budget and the existence of many other pressing priorities make the actual figure uncertain. The housing and establishment enumeration will have two principal purposes: (1) to assess the physical conditions of housing (for example, the materials used in construction), and (2) to assess congestion (number of rooms, number of persons in household, relationship to head of household). The United Nations' definition of household is used. This Census will estimate the size of the population in each area before the population Census proper is carried out and will facilitate planning for that enumeration. The population Census will go into considerably more detail than the housing Census. In addition to obtaining more information on persons living in households, it will attempt to enumerate persons without proper and permanent housing. Owing to the return of many refugees from India, this "floating" population is expected to be unusually large. The basic population Census questionnaire as now conceived will contain 28 questions, most of them on the following topics: age, marital status, children ever born, number of living children, 4

19 religion, physical disability, birthplace, nationality, number of years in place of enumeration, previous place lived (urban or rural), mother tongue, other language, literacy, education, field of education, field of training (a new item), employment status, and type of industry in which employed. For some of the questions, 100.percent of the responses will be tabulated; for others, a 10 percent sample. The census questionnaire has been designed with assistance from a Census Advisory Committee consisting of representatives from governmental agencies, research organizations, and universities. One difficulty the government anticipates is recruiting the necessary expertise for conducting the Census. Census supervisors and enumerators in the field will be recruited from the ranks of local officials and schoolteachers. About 60 percent of the enumerators will already have had experience from the 1961 Census. The Census will be preceded by a publicity campaign, since it is deemed urgent to educate respondents as well as enumerators about the purpose of the Census. The recent struggle for independence has left a reservoir of suspicion among certain segments of the population, and this suspicion must be overcome if valid answers to questions are to be obtained. Urdu speakers, for example, though very small in number, may be reluctant to report their true language status. Similarly, certain groups have special access to goods in short supply; it is anticipated that other persons will represent themselves as members of those groups unless proper precautions are taken. Another problem is that some government workers hold more than one job, which is illegal, and their fears of reprisal must be assuaged if they are to report their full labor force status accurately. A preliminary tabulation of the entire enumerated population will be prepared in the field and centrally tallied by sex, literacy, and employment status. It is expected to be issued within a month after the enumeration. A subsequent 10 percent tabulation, prepared in the central office from the original forms, is planned to be ready sometime in 1975 and a second 100 percent tabulation sometime in Some of the tabulations will be done manually, but it is hoped that for urban areas computer processing will be feasible. 5

20 1 HONG KONG Benjamin Mok, country representative Robert Retherford, rapporteur The Department of Census and Statistics conducted a de facto population and housing Census of Hong Kong in March The canvasser method was used and over 17,000 enumerators and 1,000 chief enumerators were engaged for the field work. The Census of Housing and Living Quarters gathered data from some 4 million personal records, 700,000 living-quarter records, and 900,000 household records. An efficient processing system was required, therefore, to handle the large volume of data quickly, to check and edit the inputs, and to make cross tabulations. Furthermore, the need for a variety of detailed geographical tabulations, for planning purposes as required by different government departments, dictated an extremely flexible system of tabulation. The Department of Census and Statistics decided to use the facilities of the Government Computer Centre to process the Census data. The equipment consisted of an International Computer, 1903A Central Processor with 48K words core storage, a card reader with a 300-card-per-minute capacity, a document reader (universal document transport), three exchangeable disc drives, four magnetic tape units, and two line printers. For the heavy imputation run and various tabulation runs with a number of big tables it was decided to use FILAN, a new software system called the FILE ANALYSIS SYSTEM, developed by International Computer, Ltd. (ICL). The Census data processing system consisted of four phases: (a) input, (b) organization, (c) imputation, and (d) tabulation. Input phase. The document reader was used to read Census data marked on a form called the U.D.T. form. The average reading speed was estimated at about 5,000 forms per hour, and the proportion of rejects was, on the average, about 8 percent. It took some 1200 computer hours to read the 5 million forms plus rejects. The input phase commenced at the end of March 1971 and was completed by the end of July Organization phase, This phase included the sorting of various types of records into proper sequence and checking for structural completeness, out-of-range variables, etc., and amendments of the errors detected. This phase began in mid-april 1971, soon after the Inputing work of the records of the first district was completed. A total of 400 computer hours was used for the organizational phase, which was completed on 23 August

21 Imputation phase. This phase was necessary because enumerator errors, overlooked by the field editors, were still present in the data. The work involved editing the organized data in order to remove inconsistencies, with the purpose of building into the program the ability of a human editor to check for Inconsistency and make amendments based on complimentary information. The imputation phase took about 200 hours of computer time and was completed in the first week of November Tabulation phase. District data files were read into the computer by programs and the data were compiled into tables showing the relationship between cross-tabulated characteristics. The major program, consisting of more than 150 heavily cross-tabulated tables, was carried out in four computer runs, two for personal characteristics cross-tabulated with each other, and two for living-quarter and household characteristics cross-tabulated with each other and with personal data. The processing time for the first two progarms was nearly two-thirds longer than that for the second two. This was because, when personal data were compiled into tables, each of the personal records, numbering 4 million, had to be evaluated before a value could be entered in the table cells. When living-quarter and household data were compiled into tables and cross-tabulated with personal characteristics, however, only the living-quarter and the household records (numbering 1.5 million) had to be evaluated. The total processing time for the major tabulation program was about 1,700 computer hours, and this job was completed in March Special tabulations of data on transport, street blocks, poverty, public housing, and other characteristics were also undertaken; some of these were completed in 1972 and the last run is being processed now. In response to questions, Mr. Mok stated that socioeconomic tabulations have not been made, but detailed information by occupation subgroups is available on summary tapes, which could be used for this sort of tabulation. A 1 percent sample tape has been sent to ECAFE, and a 5 or 10 percent tape is in preparation. Publication of the major Census reports, including the technical reports, is expected in two or three months. Some reports have already been issued. 7

22 INDONESIA Abdulmadjid and Sam Suharto, country representatives Robert Retherford, rapporteur In planning the 1971 census, the Central Bureau of Statistics strived for comparability with the 1961 Census, Since the earlier Census had a number of shortcomings, however, it was necessary to Introduce changes that somewhat reduced comparability. Another goal was to improve the data processing and to publish the results quickly so that they would be of maximum usefulness to government officials working on the Second Five-Year Development Plan. Among the changes In the 1971 Census was a reference period for reporting economic activities of one week, compared with one of six months in the 1961 Census. For respondents engaged in agriculture there were additional questions. One, for example, asked whether the respondent had worked during the last season; the purpose was to provide information on the problem of overpopulation in the agricultural sector. Similar questions were designed to measure seasonal shifts, for many agricultural workers are employed In the nonagricultural sector during the slack season. Two enumerations were carried out: a complete count of the population and a 3.8 percent sample count. The basic Census form contained three questions, on age, sex, and citizenship. The sample questionnaire solicited information on geographical characteristics, age at marriage, marital status, relation to head of household, education, literacy, economic characteristics, materials used to construct home, and presence of electricity in home, among others. The Census was a combination of de facto and de jure enumeration. Urban areas and villages were divided into Census blocks of about 30 to 70 households each. A special effort was made to count persons with no permanent residence. In 1968 a Census had been conducted in West Irian, and therefore the 1971 Census adopted a simplified form for use in that area. The Census forms were tabulated manually in the field and tallied at successively higher administrative levels for purposes of a preliminary count. Final processing is being handled by a FILAN computer system produced by ICI (the same system used by the Hong Kong Census). An advance tabulation consisting of 10 percent of the 3.8 percent sample was published in August Tabulations for the entire 3.8 percent sample have been delayed by a shortage of electricity in Jakarta but are expected to be ready for publication by August

23 The recommendations from the last census workshop on own-children tabulation are being carried out, and own-children tabulations are now completed for some provinces. It is expected that the balance of the tabulations will be completed by the end of In response to a question, Mr. Abdulmadjid reported that the proportion of the workforce in agriculture declined between 1961 and 1971, but that part of the decline may be an artifact of the change in reference period. JAPAN Hideshi Honda, country Robert Gardner, rapporteur representative The original returns from the 1970 Census were transcribed onto special optical sheets that were read by a Nippon Electronic Company (NEC) computer at the rate of 200 sheets per minute. There were over 65 million sheets to be read, yet fewer than 1 percent were rejected because of inadequately handled sheets. A similar NEC computer was used to process the data. Seven basic tabulations were made: a preliminary count of the population, a final count, the basic tabulation, a tabulation of the commuting population, a 1 percent sample tabulation, a 20 percent sample tabulation, and a tabulation of internal migration. Most of these reports are already published or will be available by the end of Mr. Honda answered questions about the coding of migration statistics to assure knowledge of place or origin as well as place of destination, and about the Japanese concept of Densely Inhabited Districts (DIDs). REPUBLIC OF KOREA Young Kwon Kim, country Robert Gardner, rapporteur representative A preliminary report of the 1970 Census was prepared from a manual count, which allowed its publication at an earlier date than would have been possible if the computer visual scanning had been used. The computer is being used, however, for the tabulation of the second and third stages. Of 11 provincial volumes to be published, seven have already appeared; the remaining four and the total country volume are due to be published by April In addition to further detailed volumes to follow these basic ones, there will be a series of Census monograph studies on such topics as mortality and life tables, projections, the labor force, and migration; all of these should be published by the end of March

24 MALAYSIA R. Chander, country representative Robert Gardner, rapporteur The 1970 Census of Malaysia was 100 percent complete, with no samples, and consisted of four forms: one on agriculture, one on housing, and two on population (households and Individuals). All the data had been put onto tape by January 1973, with the firstlevel tabulation already completed for four of the 13 states. The age distribution for the total population and for each major community group, by rural/urban residence, has just been released. Studies to be published in relation to the Census will include one on boundary changes and another on projections for the country as a whole and for each state. A monograph series will also be published, including such topics as socioeconomic status, fertility, and migration and urbanization. Mr. Chander answered questions about the details of the census enumeration and about whether intermarriage poses a problem of community (ethnic) grouping in Malaysia. Apparently there is little intermarriage and the problem does not arise. PAKISTAN Hafiz Sheikh, country representative Robert Gardner, rapporteur The Pakistan Census was delayed, first by internal strife and then by war, until September Moreover, there has never been a permanent Census organization before now, and statistical functions have been fragmented throughout the government. These problems are now being solved. Mr. Sheikh described the Census enumeration, problems resulting from the great variation in density and size of the four Pakistani provinces, and also the difficulties inherent in the fact that computers read in a different direction from that of the language (Urdu) used in the Census enumeration. In August 1973 a housing and demographic survey will be taken that will include questions on migration on children ever born, which were not asked in the 1972 enumeration. In addition, beginning in July 1973, Pakistan will set up a continuous population register under the auspices of the Census organization. A general discussion followyd, centering on the problem of under-enumeration. Mr. Sheikh said estimates for Pakistan had ranged between 5 percent and 10.5 percent; an independent Census Evaluation Survey is now under way, conducted by the Central 10

25 Statistical Office. Mr. Ahmad of Bangladesh noted that in 1961 some double-counting had been detected. Dr. Cho said the undercount might be about 3 percent in the Republic.of Korea, and Mr. Chander suggested percent for Malaysia, based on the Post-Enumeration Survey. Mr. Majid reported that some estimates for the totals in Indonesia were about percent. Dr. Taeuber of the United States discussed how even post-enumeration surveys fail to get at some of the sources of underenumeration. THE PHILIPPINES Hidalgo Chaves, country representative Anuri Wanglee, rapporteur The 1970 Census of the Philippines was originally scheduled to be taken in February Owing to budgetary difficulties, however, it had to be postponed to May 1970, and a prior listing of households was not undertaken as planned. Local school teachers were recruited to serve as enumerators. Two types of schedules were used in the Census, the ordinary household schedule and a sample household schedule containing 34 questions. In addition, a separate form was used for institutional households. Because trained programmers were lacking, the Bureau of the Census and Statistics used a program developed by IBM (Philippines) with the financial assistance of the United Nations to do the Census tabulations. Preliminary counts of the population, which were based on a 5 percent sample of private households, and the advance provincial and national reports have been released. The provincial series reports provide data on economic activity, migration, fertility, education, and other social and demographic characteristics classified by urban-rural residence, age, and sex. The criteria for classifying urban areas Include density, street pattern, and presence of commercial establishments. Migration data include place of birth and place of previous residence five and ten years preceding the Census. Fertility tables were also included because of the government's special interest in fertility in the Philippines. The preparation of the final reports is now under way and should be completed by mid-1973, to be followed by the national report about three months later. The final provincial report series also includes brief descriptive analyses of the data, e.g., population growth rate, sex ratios, and median age. After completing the final reports, the Bureau plans to present special detailed tables required for research. The Philippines plans to conduct another Census in

26 REPUBLIC OF CHINA Tun-yih Lu, country representative Anuri Wanglee, rapporteur Although the Republic of China usually takes a decennial Census in years ending in 6, the demand for data required for development planning made it necessary to conduct three Censuses in 1970 the Sample Census of Population and Housing, taken December 16; the Census of Industrial and Commercial Establishments; and the Agricultural and Fishing Census. The sampling fraction for the Sample Census of Population and Housing was 5 percent of total households. Nine items on population characteristics and five items on housing characteristics were asked All tabulations were completed by April 1972, and the results of the Census were printed in a final report, consisting of four volumes as follows: 1. Statistical Abstract and Description of the Census 2. Population Characteristics 3. Economic Characteristics 4. Housing Characteristics Analysis of the Census results indicates that the average annua rate of growth was 2.43 percent between 1966 and 1970, about 0.87 percent lower than for the period The decline in the 0-4 age group is attributed mainly to the government's programs of economic development and family planning. The Republic of China still faces the prospect of considerable population growth, however, because the children born during the baby boom of the 1950s are now ready for marriage and the traditional desire for a large family still prevails in the rural areas. The increase in the number of females of reproductive age and the evidence that declines in fertil ity have so far occurred mostly among women 30 years of age and over suggest the possibility of another baby boom. The young age structure of the population may also impede economic development. Lack of a permanent Census agency is a problem. Although the Republic of China has had a registration system for many years and the household registration is accurate, there is still a need for censuses. Comparison of Census and registration data has been useful in studying the completeness and accuracy of both systems. 12

27 THAILAND Ektritra Kohkongka, country Anuri Wanglee, rapporteur representative Thailand conducted its Population and Housing Census in April A complete enumeration was made of the population in all areas and of housing characteristics in the municipal areas, whereas in nonmunicipal areas data on housing were collected from a 25 percent sample of households. Two main forms were used the listing form and enumeration form. Local school teachers were recruited to serve as enumerators, and principals served as field supervisors. Technical supervisors were also assigned for the duration of the field work, which lasted about one month. At present, a preliminary report has been released that provides provincial and national counts of population by sex and number of households. Only 40 percent of the tabulation for the basic provincial series has been completed, however, owing to a one-year delay in obtaining punch machines. It is expected that the remaining basic province reports will be completed by the end of By April 1973, the basic series for the 14 provinces of the southern region and the regional report should be completed. The 1 percent sample tabulations of basic characteristics for regional and national totals should also be released within three months. Mrs. Wanglee added that the census tabulation plans consisted of three phases: (1) a manually tabulated preliminary report that was released six months after the Census was taken; (2) the basic provincial series (which provides population and housing characteristics consisting of 71 province reports), four regional reports, and ie volume providing national totals; and (3) detailed tables giving data on such characteristics as fertility, economic characteristics, and migration. These are to be tabulated from sample data that are now being edited and coded for additional information on fertility and family composition. Tabulation of the third phase will not start, until the basic series is completed. U.S. CENSUS TABULATION PROGRAM Conrad Taeuber, country representative Robert Gardner, rapporteur The Census of the United States is one of the oldest regularly occurring censuses in the world. The 1970 Census consisted of a 100 percent tabulation of basic questions needed to fulfill the 13

28 Constitutional requirement for a decennial headcount, and 15 percent and 5 percent samples to obtain more detailed information. The 100 percent enumeration was self-coding, so that tabulation could proceed at once; but this was not true of some of the sample questions, such as those on migration. The use of preliminary reports and their publication allowed many local problems to appear and be corrected before the final reports were published. Four series on the population count were published: PC1-A, giving numbers for different areas; PC1-B, giving age-race-sex-marital status breakdowns; PC1-C, from the samples, giving general social and economic characteristics; and PC1-D, giving detailed cross tabulations from the samples. Other "publications" by the Bureau of the Census are numerous. A tape program, for example, issues tapes containing all of the data from the printed volumes and more; these tapes were actually available before the printed volumes. A series of evaluation reports deals with the problems and quality of the Census, including a series of post-enumeration studies and matching studies. A series of procedural reports has also been published, describing the actual collection and tabulation activities. Dr. Taeuber described the Public Use Sample, which consists of tapes containing information from 1 percent of the original returns and allowing cross tabulations and manipulations not presented in the published tables or other tapes. He noted several new emphases in the 1970 Census: small-area data were increased, there was more information on families and households, and sources of income were more detailed. Planning for the 1980 Census will begin on July 1,

29 DISCUSSION COMPUTER PROBLEMS IN TABULATION PROGRAMS Edmund Murphy, Robert Gardner, chairperson rapporteur Dr. Murphy spoke about computer problems in connection with the Census of Canada. He said that perhaps the single most important problem was associated with editing and imputation for the tabulations. Dr. Suharto noted that editing was also a major problem in Indonesia. Several other participants commented that testing the edit programs on small samples did not turn up all the problems, and that many did not appear until the main body of data was run. Miss Fernandez added that careful and logical programming of the edit programs.before they are run just cannot anticipate all of the possible problems. A discussion then ensued on the problem of imputation and allocation in censuses when data are missing for individual returns. Some countries publish their imputation procedures and percentages, others do not. Dr. Taeuber noted that the United States Census volumes have an appendix that deals with this problem and lists imputation percentages. Dr. Hauser suggested three levels of consideration in dealing with the imputation: (1) Is the imputation merely "cosmetic," or does it lead to statistically significant manipulations? (2) There are ethical considerations: are statisticians morally bound to report all of their techniques and practices? (3) There are political considerations, and to some extent the census can use discretion as to where and how any publication on allocations is made available. Dr. Cho opened a short discussion on the question of whether the participating countries preferred packaged programs for their census tabulations to writing their own programs. Mr. Chander stated that his organization would like to build up its own programs, but owing to limited resources and trained manpower, the packaged programs are quite useful. Mrs. Wanglee noted that in Thailand the Census organization had developed its own programs before the packaged ones became available. Mr. Mok said that the packaged programs will probably be used again in the next Census of Hong Kong. 15

30 ;» TABULATIONS ON RESIDENCE Dorothy Fernandez, chairperson Sam Baum, rapporteur The participants discussed national practices regarding definitions of urban and rural and the use of small-area data tabulations. The point was made that data on very small areas are useful as build ing blocks for the construction of statistics for special areas that are not normally tabulated in the general census tabulations. Planners often require data for such special areas as school districts, transportation areas, fire districts, and health districts. Dr. Cho asked whether the definitions of urban used in the latest censuses presented problems of comparability over time. Also, did the census definition of urban correspond closely enough the the real situation? Mr. Honda stated that Japan uses the "Densely Inhabited Area" concept, which examines population density and characteristics of the population within an area so as to include persons living outside of the political boundaries of a city who are actually urban dwellers and exclude persons living within a city who are actually rural dwellers. Owing to the lack of international comparability in urban and metropolitan area definitions, it was noted that more attention should be paid to these problems in the 1980 censuses. Several participants Indicated that they had been developing definitions of urban areas and related concepts (e.g., urbanized areas) for the purpose of giving information beyond that provided by political definitions of city populations. Mr. Sheikh mentioned that Pakistan has the concept "Urban Areas for Census Purposes," which uses socioeconomic characteristics of the population in an area as well as size and density as criteria. Mr. Kim reported that in 1970 the Republic of Korea also used such criteria as percentage of the population in nonagricultural occupations and percentage of houses with electricity to define urban areas. The 1966 Korean Census data will be retabulated according to the new definition. Dr. Hauser pointed out that even if areas are not originally treated as urban in census tabulations, regrouping of areas along some continuum of modernity can be useful in analytical work where the relation of urbanization to other variables (e.g., fertility) is under investigation. Mr. Ahmad and Mr. Sheikh both expressed the view that developing countries will need to establish their own criteria for defining urban areas, since rural areas are becoming quite densely populated and agricultural villages are being electrified. 16

31 TABULATIONS ON HOUSEHOLD, FAMILY, MARRIAGE, AND OF HOUSEHOLD RELATIONSHIP TO HEAD Dorothy Fernandez, Sam Baum, rapporteur chairperson The session was opened by Dr. Cho, who asked the other participants whether in their own countries tabulations by household were more useful than tabulations by family, or vice versa. Dr. Palmore pointed out that poor definitions of family in census tabulations often cloud analyses of family change. Dr. Hauser reported that the Organization of Demographic Associates (ODA) has developed a family classification framework for use in Southeast Asia that serves analytical purposes better than the standard classifications used in most censuses. The essential elements of the ODA classification are the following categories: (1) single individual, (2) nuclear family, (3) polynuclear family, horizontal extension, and (A) polynuclear family, vertical extension. Singapore attempted to use a similar classification in its Census and also tried to use retrospective questions to get at changes in family structure. Currently, use of the ODA classification requires a manual coding operation. The question arose whether countries in the region follow the U.N. recommendation of obtaining data on the number of nuclear families in a household. Mr. Chavez said that the Philippines does, Mr. Sheikh said that Pakistan does not. Mr. Honda stated that Japan follows the U.N. recommendation and in addition produces tabulations that distinguish one-generation, two-generation, and three-generation households. Dr. Taeuber asked whether the fact that many censuses in the region were conducted on a de facto basis caused problems in tabulating family and household structure. According to Miss Fernandez, although the regular Malaysian Census tabulations were confined to family members currently living at home, information from the household record could be used to reconstruct families if members did not sleep in the household on the night of the Census. Mr. Kim reported that the Republic of Korea classifies families by agricultural or nonagricultural type, and Mr. Honda said that in Japan household data classified by economic type are useful in labor force analysis. Dr. Cho informed the group that if relationship to head of household is obtained for each individual, fertility estimates can be prepared by the own-children method. He asked whether it was possible to determine mother-child relationships in each country's census schedule. The major problem arises, he noted, when two or more mothers reside in the same household. Mr. Mok raised the 17

32 the question of how household head is determined. In Indonesia the major earner is the head of household, Mr. Abdulmadjid said. Mr. Honda stated that although'the major earner is also considered to be the household head in Japan, in the countryside the oldest male may be named as the head even if he is not the major earner. Mr. Sheikh pointed out that different definitions of household head may be useful depending on the type of analysis being undertaken. Mrs. Wanglee of Thailand and Mr. Kim of the Republic of Korea both indicated that in their countries the head of household is determined by self-identification. Several other participants reported that in their countries the husband is always treated as the head of household in their editing procedures. On the question of identifying the mother-child relationship, Mr. Sheikh noted that in the Pakistan Census schedule children are identified by the mother. In Indonesia, Mr. Suharto said, the Census instructions prescribe listing of members in a specified order that permits identification of the mother-child relationship; in about 10 percent of the schedules, however, these instructions were not followed. A special tape is being prepared in Indonesia for households with only one mother present. For the remaining households with more than one mother present, a computer printout will be made that can be manually coded to identify the mother-child relationships. Mr. Lee reported that in Hong Kong tabulations of own children can be made only for nuclear households. Mr. Chavez stated that in the Philippines identification of own children was possible only in onemother households. TABULATIONS ON FERTILITY AND MORTALITY R. Chander, chairperson Fred Arnold, rapporteur Some basic questions on fertility appear to be universal in the censuses of countries participating in the workshop, but the precise questions asked and the analytical usefulness of the data collected vary somewhat across countries. In most of the countries no data were collected directly on mortality, although mortality can often be inferred by manipulating basic data on fertility and household composition. The most common fertility questions on censuses sought information on the number of children ever born to women in each household and the number of children currently living. 18

33 Dr. Cho suggested that, in the absence of adequate vital registration statistics, census data can be used to estimate fertility and mortality. The method of estimation depends on the tabulation of children ever born by single years of age and the tabulation of own children living by single ages of the children and the mother. On the basis of these tabulations, it is possible to arrive at an estimate of child mortality from which one can infer mortality at other ages. These estimates in turn lead to an estimate of agespecific fertility rates. Dr. Cho further suggested that- the publication of a table similar to the one reproduced below would greatly facilitate the estimation procedure. Ages of women in single years All women Evermarried women Number of children ever born Number of living children Children living Total <1 1 2 > V 60 unknown It would be useful, whenever possible, to separate the unknown category into three groups: (1) mothers not living, (2) mothers living away from children, and (3) mothers whose age is unknown. Finally, even if the age distribution of children is subject to age heaping, the adverse effects can be attenuated by combining some age categories and by truncating the series, if necessary. Dr. Cho also emphasized the utility of calculating these estimates separately for different socioeconomic groups in order to gain some insight into differentials in mortality by sex. Several participants indicated that the necessary data for the above calculations were either not included in their censuses or else were lost at the coding stage. The major problem was an inability to match children with their own mothers in households with 19

34 non-nuclear family structures. When this situation occurs, it is possible to make fertility estimates on the basis of nuclear families alone. Another problem arises when children are not living with their mothers. Although this problem may be minimal for many Asian countries, such as Malaysia and the Republic of Korea, Dr. Taeuber pointed out the difficulties presented by black families in the United States where a large proportion of children are not living with their mothers. Mr. Baum suggested that the own-children method of estimating fertility and mortality might be based on sample tabulations, and Dr. Cho agreed that this would be feasible for the national level. Descriptions of the data collected in several of the countries helped to bring out both the differences from one country to another and the elements that were common to most countries. The Malaysian Census collected data on the number of children ever born, the number of living children and whether they were living at home or elsewhere, the number of children who had died, and the number of stillbirths. Other countries avoided these last two questions because of their sensitive nature. Mr. Lu reported that the Republic of China collected data on the total number of children delivered, the number of living children, and the number of children living with their mother on the 1966 Census. Mr. Kim mentioned that tabulations of number of children ever born and number living by age of wife would be published for the Korean Census; the same data were tabulated from the 1966 Census. Mr. Chaves noted that similar information was available for the Philippines except in cases where three or more families were included in the same household. In Hong Kong age-specific fertility rates were calculated by the own-children method for nuclear families but in non-nuclear families it was not possible to relate the age of children to the age of their mothers. Dr. Murphy reported a relative lack of interest in general fertility research in Canada (possibly because of the sparse settlement In the country) but considerable interest in French-English fertility differentials. Mr. Ahmad reported that Bangladesh plans to collect data during the forthcoming Census on the number of children ever born (excluding stillbirths) for ever-married women. Dr. Taeuber discussed efforts to develop measures of childspacing by interpolating to estimate whether children no longer living at home died or whether they are living elsewhere. The session ended with a discussion, initiated by Dr. Taeuber, of preliminary plans for the World Fertility Survey and its potentialities in Asia. 20

35 TABULATIONS ON ETHNICITY, RACE, AND MINORITY GROUP STATUS- R. Chander, chairperson Fred Arnold, rapporteur The extent of concern with matters of ethnicity, race, and minority group status varies considerably from one country to another, and each country has to deal individually with its own unique problems in this area. The decision as to what questions on this topic should be included in censuses is often made more on the basis of political pressures than on the basis of legitimate research needs. Although some countries have relatively homogeneous populations and can therefore avoid this problem area, in other countries there are crucial distinctions to be made between various ethnic groups, religious groups, or even language groups. Mr. Chander reported, for example, that in Malaysia all tabulations from the Census are controlled by the four major ethnic groups. Mr. Sheikh stated that since more than 98 percent of the population of Pakistan is-muslim, only a few tables deal with religion; the volatile subject of language was avoided altogether in the last Census. Mr. Ahmad noted that there are no problems with mother tongue in Bangladesh because of the homogeneity of the population, but a question on religion is planned for the next Census. In the Hong Kong Census, according to Messrs. Mok and Lee, there is no question on race but extensive tabulations based on place of origin are published. The Indonesian Census contained no questions on ethnicity, again because of the relative homogeneity of the population. Questions on citizenship and religion were included in the Census, however. Dr. Murphy emphasized the divisive effects of language differences in Canada and said that three separate language questions were included in the Canadian Census: (1) mother tongue, (2) official languages respondent Is able to speak, and (3) language spoken most often at home. He also mentioned that some problems had arisen with questions of ethnic origin because of changing boundaries in eastern Europe. Mr. Honda reiterated the ethnic uniformity of the Japanese population. The Japanese Census included a question on nationality, but fewer than 1 percent of Japanese respondents are not nationals and the only major tabulations on nationality are by age and sex. Mrs. Wanglee reported that the Census of Thailand included a question on religion but that ethnicity is sufficiently homogeneous to preclude the necessity of additional questions on minority groups. Dr. Taeuber stated that the U. S. Census includes questions on country of birth of respondent, country of birth of respondent's parents, race, and mother tongue; he indicated that these data will be extensively utilized and that seven separate subject reports will focus on ethnicity. He also noted that neither the 1970 Census nor any previous U. S. Census included a question on religion. 21

36 TABULATIONS ON MIGRATION Sam Suharto, chairperson Geoffrey McNicoll, rapporteur Dr. Suharto described the treatment of migration in the 1971 Indonesian Census. Four questions were asked: (1) place of birth; (2) whether the person had ever lived in a different province; (3) if so, the name of that province; and (4) the length of time lived in the present province. Imputation rules for transcribing the answers had to be designed with care. For example, it was common for people to have been born in another province without having "lived" there. The Census did not ask for rural or urban origins of migrants and thus could not provide data on rural-urban migration. One reason this was not done, Mr. Abdulmadjid pointed out, was that many place names were duplicated and the name alone was insufficient to identify a place as urban or rural. Mr. Mok said that the Hong Kong Census did not enquire about internal migration. A question on place of birth elicited information on external migration, and it was found that one-half of the population had been bom on the mainland. Information was also sought on commuting habits of the work force. A sample Census in 1976 will further investigate internal movements for housing and town-planning purposes. Mr. Chander listed the migration questions in the Malaysian Census: place of birth (state); for those born overseas, length of residence in Malaysia; length of residence in present place; where last resided (town/other). The last question elicited from the respondent a subjective determination of whether he had previously resided in a town or other place. Movement of persons between West Malaysia and Singapore, prior to 1971, made estimation of this component of migration difficult. It is expected that the Singapore Census tabulations will provide data on the Malaysian-born population residing in Singapore. Miss Fernandez described plans of the Department of Statistics, Malaysia, to tabulate the data on migration. The plans include tables on inter-district migration with urban-rural breakdowns, and cross tabulations by socioeconomic status. A special monograph on the subject Is being prepared. Mr. Honda gave details of the 1970 Japanese enumeration. Movements were separated into those made in the previous five years, and those made prior to Commuting habits of the work force 22

37 and students were also Investigated. The detailed tabulations of migrant characteristics by small administrative areas have not been published but are available in the Bureau and were also sent to the respective regional authorities. The Thai Census included questions on place of birth, moves in the previous five years, duration of stay in present residence, province of last residence, and for migrants whether they moved from a village or a municipal area. Mrs. Wanglee noted that although the data include only moves between provinces, Thailand has 71 provinces. Cross tabulations will be made by age and sex. Mr. Chavez reported that in the Philippine Census questions on migration included place of birth by province, and if born in the same province whether or not born in the same city or town; and place of residence five and ten years prior to the Census. Cross tabulations will be made by age group only; the Census will not yield data on rural-urban migration. Mr. Lu said that in the Republic of China migration statistics are derived from registration data. The remaining discussion dealt with general issues rather than the experience of particular countries. Dr. Murphy pointed out the conceptual difference between a question on place of living five years ago and one on place of last residence. The former had the conceptual neatness of a fixed reference period and fitted well into formal demographic analysis. The latter, which he had come to prefer since leaving academic life to become a practicing statistician, gave a better picture of migration patterns on a year-to-year basis. He was interested to know why particular countries adopted one or the other approach. Dr. Hauser argued for collecting both kinds of migration information. Difficulties of obtaining information on rural-urban migration were discussed at length. Miss Fernandez described the procedure used in Malaysia for studying rural-urban migration. It was suggested that a single definition of urban and of rural would eliminate the possibility of confusion caused by.the two sets of definitions currently used. As an alternative, Dr. Hauser suggested grouping administrative districts by some composite measure of the degree of urbanization, for example by using the percent of nonfarm employment. For the purpose of making international comparisons of urban migration, he noted that it might be desirable to tabulate the characteristics of migrants to the largest cities in each country. Dr. Hauser described the U.S. Census concept of "State Economic Area" (SEA). SEAs are functional units within a state, selected on the basis of "nodality" (combining a city with its hinterland) and 23

38 economic or ecological homogeneity (for groups of counties). Studying migration between SEAs can give substantially more insight into underlying patterns than looking at migration between standard administrative units. Broader ecological divisions, cutting across state boundaries, are also used in the United States; nine such regions are identified (e.g., "coastal plains"). Dr. Hauser strongly urged census offices in Asia to anticipate future requirements for data classified by natural or functional units. Where there are rigid administrative divisions in a nation that would make the identification of overlapping ecological regions a sensitive political Issue, smaller areas similar to the SEAs could be used instead. TABULATIONS ON EDUCATION Sam Suharto, chairperson Geoffrey McNicoll, rapporteur The discussions in this session ranged over three topics: (1) specific questions asked in the recent censuses, including the definitions used; (2) the quality of census educational data; and (3) plans for tabulation, analysis, and dissemination of the results. Census questions on education All countries had asked questions on educational attainment, usually number of years of school completed. Dr. Taeuber noted that in the United States there was also a question on highest grade attended. Vocational schools are in general treated in the same way as regular schools, although they are classified separately in several countries. Questions on field of study differed substantially. In the United States, the question was asked only of those who had completed a vocational program; in Malaysia, Indonesia, and the Philippines, information on broad field of study was obtained; in Hong Kong the question was not included. Religious schools pose various problems. Mr. Sheikh noted the existence in Pakistan of Islamic schools where students learn to recite the Koran without necessarily learning to read and write. Even if the students attain literacy, their education is often limited to religious subjects. Mr. Abdulmadjid said that government pressure has induced similar schools in Indonesia to move toward a broader curriculum; purely religious schools s t i l l exist, but mainly at a higher level. Mrs. Wanglee described the procedure used in the Thailand Census for relating grades in religious schools to the general school system. 24

39 Other matters discussed more briefly included the lower age limit for the question on education (variously Oi 5, 10, and 12 years) and difficulties in determining literacy. Quality of census education data Several participants noted the distinction between educational attainment and any objective measure of achievement. Mr. Schmittdescribed how-the average education of a cohort as indicated in succeeding censuses tends to improve as it ages. Mr. Sheikh pointed to a bias in the Pakistan Census data caused by interviewers 1 reluctance to use the longer form required for respondents over ten years of age. It was observed that such problems are common elsewhere. Mr. Abdulmadjid noted a difficulty that has resulted from the small sampling fraction in the Indonesian Sample Census: educational data were obtained from the Sample Census only and do not give accurate results for the small numbers of persons with high educational attainment. Disparities are often found between school attendance data derived from the census and independent estimates of school attendances made by ministries of education. The latter agencies are sometimes critical or skeptical of the census figures. Mr. Baum emphasized the importance of evaluating the quality of the census in order to counteract such criticism when it is unjustified. Mr.'Mok said that in Hong Kong this disparity is not found and that school enrollment data compare well with the Census. Tabulation, analysis, and dissemination.in contrasting census data with typical school enrollment statistics from education ministries, Dr. Hauser stressed that the chief advantage of the census is that it can provide cross tabulations of educational attainment with other characteristics. Elaborate cross tabulations available or planned were reported by the participants from Malaysia, Hong Kong, the Republic of Korea, and Indonesia. In the United States, Canada, and Malaysia, samples of educational and occupational.groups drawn from census data have been studied in detail to provide additional information on educational background. Mr. Baum pointed to the need for educating consumers, including in some instances ministries of education, on the uses of the census tabulations. Among the many uses discussed in the session were the following: 25

40 Allocation of resources within the educational sector (both to correct present misallocations and to alert planning agencies to future allocation problems) Studying invisible underemployment (mismatching of training and occupation) as part of a larger investigation of the effectiveness of labor utilization Planning the eradication of "functional Illiteracy" (defined in the United States as less than four years of school) Studying the characteristics of school drop-outs Monitoring the acceleration or retardation of children's progress through school grades, both over time and between regional,or other groupings. TABULATIONS ON OCCUPATION, INDUSTRY, CLASS OF WORKER, INCOME, AND LABOR FORCE PARTICIPATION Bahauddin Ahmad, chairperson Philip M. Hauser, rapporteur At the invitation of the chairperson, Dr. Stephen Yeh summarized the census situation in Singapore. The results of the 1971 Census will be published in two volumes which are in the final stages of publication for public use. Volume 1 contains the Census tables, a list of which is now available. Volume 2 contains brief Census analyses and other statistical data including population trends. The committee that helped to plan the Census made provisions for three sample studies (10 percent samples) of (1) the labor force in relation to education, (2) fertility, and (3) housing conditions. Program cutbacks, however, will result in less information than planned on the labor force and housing, although data will be available on fertility. Discussion then focused on problems and Issues of comparability in the tabulations on occupation, industry status, and labor force participation. Mr. Ahmad began by pointing to problems of occupational classification and the frequent need for countries to obtain more detail on some occupations than is provided for in the Standard International Occupational Classification. He devoted attention to the difficulties of coding occupations and to problems of quality control arising from inadequate occupational entries on the schedule. Mr. Ahmad then called upon the representatives of the various countries to summarize their experience in producing labor force statistics. 26

41 Japan Mr. Honda reported that Japan follows the International Classification but experiences difficulties arising from some inadequate occupational returns. In.certain cases, communication is necessary for the field to obtain adequate occupational descriptions, especially in the large cities. Clearance with the field does not introduce great delays in processing because editing and coding operations continue. Canada Dr. Murphy reported that Canada used a three-digit occupational code and made special efforts to obtain detailed occupational descriptions. Coding was carried out in regional offices largely by college students who were trained for i t. There is more than one source for labor force.statistics, however, and inconsistencies exist in the data from the various sources. (The Labor Force Survey results did not agree with the Census results, for example.) Detailed labor force tables were tabulated from the Censuses of 1961 and 1971 that give occupation by. industrial class of worker and by age and sex. Indonesia Mr. Abdulmadjid reported that his country plans to use a threedigit occupational classification but will report on a two-digit basis. The Standard Classification has been expanded to obtain needed national data. The schedule calls for Information on what the person does, where he does i t, and what the product is. Problems include those arising from different local languages and different names for the same occupation. Hong Kong Mr. Lee reported that questions on occupation, industry, and class of worker (activity status) were asked of all persons and coded by the enumerators themselves using a two-digit classification. Hong Kong is planning for a three-digit level in its Census of 1976, at which time they hope to use a team of trained coders. Some of the occupational data cross-classified by the two-digit industrial data. Extensive tabulations have been made with occupation crossclassified with other data such as education. The tabulations are adequate for most uses. Republic of Korea Mr. Kim reported that it is difficult for enumerators with limited experience to get good occupational data. It has been necessary to give intensive training to obtain good results. 27

42 In response to a question, Dr. Hauser spoke of the problems of occupational and industrial classification. He pointed out that censuses could not possibly use the detailed classification system of the type required for job placement, as manifested in the U. S. Dictionary of Occupational Titles. Most of the occupational returns obtained from a population would be different than those obtained from the establishment approach. It is necessary for the census to recognize its limitations and to develop the best possible data within them. Dr. Taeuber reported on thetapproach used in the U. S. Census for obtaining occupational information and read the questions contained in the 1970 schedule. He reported that the occupational coding was performed by a crew of trained coders and constituted a large part of the manual processing. Malaysia Miss Fernandez reported that a three-digit occupational classification was used in Malaysia but that only a two-digit classification system will be published. The Malaysian Census did not obtain income information but gathered data on household equipment; plans are to construct a socioeconomic index based on the household equipment, the quality of housing, and occupational rating. During a discussion about the possible use of an occupational rating, Dr. Hauser encouraged experimentation and Dr. Palmore reported a lack of success in his efforts to rate occupations in Malaysia due to the differences in ethnic occupational experience. United Nations Mr. Yu questioned the representatives of Malaysia, Pakistan, and Indonesia about secondary occupations. All three reported that secondary occupational information was not obtained in their most recent Censuses. Mr. Sheikh asked how labor force information could be presented for only the civilian population if it was necessary to conceal the size of the military establishment. One possibility, it was suggested, is to report the specific occupations of the military that can be distributed throughout the occupational data. The Pakistan Census includes 90,000 prisoners of war s t i l l held by India. The Philippines The Philippine Census included a 5 percent sample to achieve comparability with the results of the Labor Force Survey. It reported on the economically active and inactive populations, on employment and unemployment, and also on occupation, industry, class of worker, and hours of work. The age cutoff for the labor force was ten years. Income information was not obtained because experience indicated that 28

43 the low-income groups tend to inflate their returns and the highincome groups tend to deflate theirs. Income information will be obtained by means of sample survey that will employ better trained enumerators. Although occupations have been coded with a threedigit code, only two digits will be used in the final tabulations along with a two-digit industrial code. Republic of China Mr. Lu reported that the 1970 Census obtained occupational and industrial data and used a two-digit classification scheme. It did not gather information on income. Thailand The 1970 Census collected information on type of work for persons 11 years of age and over. A two-digit occupational classification was used. Data will be presented on a one-digit basis for most of the provinces but on a two-digit basis for the larger provinces. The basic tables will show occupational data, industry, and work status by age and sex. Possibly a tabulation will be made by occupation and industry groups. Pakistan Information on the labor force was collected in the Censuses of 1951 and 1961 and also for the Quarterly Survey of Current Economic Conditions, in which labor force activity was the main topic. In connection with the 1972 Census, data on the labor force are being collected in the Housing, Economic, and Demographic Survey of Questions on subsidiary occupations are being dropped because of conceptual and response problems. It is felt that the labor force participation approach is inadequate inasmuch as the statistics derived from this approach do not reflect real rates of unemployment and underemployment in the developing countries. General Discussion Mr. Honda reported that in Japan the classification of labor force status departs from.u.n. recommendations. All unpaid family workers are included in the labor force because hours of work were not obtained in the Census. Dr. Taeuber pointed out that in the United States an arbitrary cutoff point is used and that unpaid family workers are not included in the labor force unless they work 15 hours a week or more. He reported a growing interest in the labor force participation of women who are mothers with children.of various ages, and in the income and family status of households with two or more workers. Similarly there is growing interest in labor force information on youth especially urban black male youths in mobility 29

44 in labor force participation. Dr. Hauser called attention to the problems of comparability, particularly with respect to the marginal groups In the labor force the youth, the elderly, and women. He called attention to variations in definitions of who is to be included and excluded from the labor force. Mr. Chander reported that in Malaysia the standard labor force approach with the one-week reference period was supplemented by a 12-month reference period in order to deal with the problem of seasonality. He raised questions about the inclusion and exclusion of unpaid family workers. Mr. Baum emphasized the point previously made by Dr. Taeuber about the limitations of censuses, remarking that types of information must be obtained from sources other than the census. He called attention to various surveys including longitudinal studies of the work force in the United States. Mr. Chander pointed out that, unfortunately, such special studies often cannot be pursued in the developing countries because of limited resources. The chairperson inquired about variations in the lower legal age limit for participation in the labor force. Discussion indicated that the lower limit varies from ten to 16 years. Toward the close of the meeting Dr. Hauser described two new frameworks being developed for measuring the work force and its utilization by the Council for Asian Manpower Studies (CAMS) and the Organization of Demographic Associates (ODA). The,! ODA-CAMS framework," which resulted from an ODA Labor Force Workshop, is an alternative to the labor force approach. The "labor utilization framework" designed by Dr. Hauser can be used with either the labor force approach or the proposed new ODA-CAMS approach. The ODA-CAMS approach differs from the labor force approach in three respects. First, in an effort to be more realistic, it involves placing a person in a three-dimensional space in which one axis is employment In a family enterprise or a nonfamily enterprise, the second axis is employment in agriculture or nonagriculture, and the third axis is employment in the monetary or nonmonetary sector. Second, unlike the labor force approach the ODA-CAMS approach recognizes multiple employment. Third, the ODA-CAMS approach places great emphasis on the household and family as the unit for labor force tabulation and analysis. Dr. Hauser indicated that in the following session he would describe the labor utilization framework and provide the preliminary results of experiments conducted in three countries. 30

45 PROPOSAL FOR STUDY OF LABOR FORCE UTILIZATION M.H.D. Hafiz Sheikh, chairperson Philip M.. Hauser, rapporteur Dr. Hauser outlined the framework he has developed for the measurement of labor utilization using the labor force ("active population") approach for the measurement of the work force. He stated that the standard tabulations usually obtained in census or periodic surveys of the labor force report only the unemployment of the work force and fail to measure visible or invisible underemployment. Since underemployment is often the major problem in labor utilization in developing areas, there is a special need for measuring the phenomenon. The labor utilization framework involves the. tabulation of the following functional categories: (1) the total work force, (2) work force adequately utilized, and (3) work force underutilized (a) by unemployment, (b) by input (as measured by less than fulltime work), (c) by productivity (measured by low income or proxy therefor), or (d) by mismatch of occupation and education. In addition it is possible to measure the "passive unemployed" or "discouraged workers" those who would accept employment if it were available but who are not seeking work because they believe none to be available. Materials were distributed including two articles written by Dr. Hauser and tentative schedules and tabulation outlines for using the labor utilization framework either with the labor force approach or the alternative ODA-CAMS approach previously described. The materials distributed also included specifications for the tabulations to be run. Mrs. Choe described the computer program she has developed for the labor utilization framework. Census offices can obtain from the East-West Population Institute the necessary programs with approximately 200 punched cards. The programs will be prepared for use on both large and small computers. Dr. Hauser presented the empirical results of preliminary tabulations using the labor utilization framework in the Philippines, Singapore, and Malaysia. He set forth principles for determining the cutting points in hours of work to measure underutilization by input, and in income to measure underutilization by productivity. The problem of developing a code for determining the mismatch of occupation in relation to education was also discussed. It was generally agreed that the labor utilization framework provides much more information about the work force than do the standard labor force tabulations. Dr. Hauser pointed out that for each of the 31

46 functional categories in the labor utilization framework it is possible to obtain worker characteristics of age, sex, relationship to head of household, occupation, industry, class of worker, etc. that is, all the items on the schedule. TABULATIONS ON SOCIOECONOMIC STATUS M.H.D. Hafiz Sheikh, chairperson Philip M. Hauser, rapporteur Mr. Sheikh initiated the discussion of tabulation of socioeconomic status (SES) by calling attention to conflicts in criteria for determining SES. Examples are smugglers, who have high economic status but low social status, and government servants in some countries, who have high social status and prestige but low economic status. Dr. Palmore called attention to the problems of obtaining SES tabulations in developing countries because most classification systems originated in the West and have been largely inapplicable in As ia. Dr. Hauser recalled Max Weber's consideration of the problem of socioeconomic status, pointing out that three different axes are involved: (1) births, as in the case of aristocracies, (2) economic power, and (3) prestige, as associated with education or government service in some countries. Obtaining an index of SES involves in some way the simultaneous use of a l l three axes, or, conceivably, the use of different indexes based on the respective axes. Dr. Taueber outlined the work of the U.S. Bureau of the Census in developing an index of SES based on occupation, education, and income. Two measures are involved: (1) measurement of status utilizing these data and (2) measurement of discrepancies among the measures of each item. The calculations and the meaning of the index are s t i l l controversial in the United States. Mr. Chander discussed problems in measuring SES in Malaysia. There are great differences among ethnic communities and between urban and rural areas in occupation composition, education, and income. The problems of obtaining a single-dimension measurement of SES is therefore very difficult if not impossible. Dr. Hauser pointed out, however, that even rough measures of SES can provide important information and be useful as controls in tabulations of differential fertility and mortality, for example. Mr. Yu said that in his judgment no one scale of SES Is possible; 32

47 that it is important for nations to develop SES indexes for their own use; and that international comparability, although it may be possible, is less important than the national usage. Mr. Honda pointed out that in Japan SES is based on the threedigit occupational classification. The Japanese Census is concerned primarily with economic activity as a major measure of SES. This works reasonably well because Japan is a relatively homogeneous population. The occupational classification system, however, could be supplemented by other data such as income. In Pakistan and other developing countries, Mr. Ahmad stated, there are different axes for the measurement of SES including housing, education, and occupation, but income data generally are not available. Dr. Hauser noted that Paul Siegel of the University of Michigan demonstrated in his studies that SES is essentially the same whether prestige or economic criteria are used. Mr. Abdulmadjid reported that in rural areas of Indonesia housing quality is so uniform that it is impossible to obtain differentiation of status by using housing criteria. Mr. Kim suggested that proxies for income could be used in efforts to obtain indices of SES. Mr. Sheikh emphasized that cultural differences result in quite different prestige scalings and that prestige can change over time. In Pakistan, for example, the civil servant at one time enjoyed very high status but over time became subordinate to the army and, more recently, to party affiliation. Problems of comparability exist therefore even within a nation over time. Dr. Murphy called attention to political sensitivities in Canada to SES classification, which is often interpreted pejoratively. He said it would be better, therefore, to present occupational classes without reference to SES. Mr. Sheikh stated that Hindu influence in Pakistan society is manifest in differentiations of status depending on birth but that this influence is declining. In Mr. Chander's judgment, although occupational classification for SES may be satisfactory in Canada, it is not so in Malaysia; ethnic differences by occupation create distortions when occupational classification alone is used. Dr. Hauser pointed out that the index of SES must be related to those objectives of its use. It can be used as a control to obtain differentials, as in fertility and mortality, but it can also be used as a basis for fixing government policy based on differential opportunity and status. Miss Fernandez suggested that scaling of SES can vary for people with different viewpoints, and that it may be necessary to have indexes representing several viewpoints. Mr. Lu commented that housing information provided a possible basis for SES classification in Taiwan. Discussing their efforts to 33

48 develop an SES index on the basis of a Gutman scale in Malaysia, Mr. Chander and Miss Fernandez cautioned that different scales might be necessary for urban and rural communities or for the various ethnic communities. Mr. Ahmad pointed to the possibility of using land tenure as a basis for SES; ownership, rental, some combination of ownership and rental, or agricultural labor status could provide a useful basis for SES classification. Mrs. Wanglee stated that in Thailand the study of income levels is just beginning, and consideration is being given to the possible use of proxies for income. She suggested that the problem of SES in Thailand might be easier to solve than in some other countries because of the relatively homogeneous character of the population. Dr. Cho raised the question of whether countries could experiment, even during the census tabulation period, with a common SES scale if such a scale were developed. The country representatives responded as follows: Malaysia: Possibly, if the scale were made available soon enough. Pakistan: They would like to try. The SES tabulation might be based on a 2 percent rural sample and a 5 percent urban sample. Bangladesh: Such a tabulation might well follow the main Census tabulations. Philippines: They are interested in the SES tabulation but do not have a classification system worked out at present. Indonesia: If an SES classification system were developed they would be willing to try i t. Hong Kong: They probably could not do an SES tabulation of 100 percent of the population but perhaps could use a 1 percent sample. PROPOSED PROGRAM OF CENSUS DATA DISSEMINATION Sam Baum, chairperson Philip M. Hauser, rapporteur Dr. Murphy discussed methods and problems of disseminating census data, stressing the use of the computer as a means of expanding 34

49 dissemination possibilities. He noted that ingenuity in the use of the increased memory of computers, provision for data retrieval, and programs to make possible recombinations and special computations of original responses, with safeguards for the preservation of confidentiality, have greatly increased the uses of census data and made them more broadly available to the public. Other important devices for effecting more widespread dissemination of census data are the developments in making summary tapes available and machine-readable original responses, which safeguard individual returns. PROPOSAL FOR A CENSUS NEWSLETTER Sam Baum, Sandra Ward, chairperson rapporteur Dr. Cho proposed that the East-West Population Institute serve as a clearing house for census information and establish a quarterly census newsletter to facilitate continuing communication among census organizations in Asia and the United States regarding their operations. ECAFE puts out a bulletin that includes items on censuses, but there is usually a considerable time lag between an event and its publica tion and census news is only incidental to other kinds of information contained in the bulletin. Dr. Cho suggested that as census or statistical agencies issue reports, they send the Institute a copy to be abstracted for the newsletter. He also proposed that the first issue be devoted to a review of each country's progress in tabulating and publishing results of its latest census and that there be a regular section for presenting abstracts of publications on census tabulation. Dr. Hauser stressed the importance of structuring questionnaires to newsletter contributors in order to receive prompt answers. It was agreed that a quarterly newsletter would be a useful publication to those present but that participation should not be limited to those countries that had sent representatives to the Workshop-Conference. Next, Dr. Cho proposed that the census agencies represented at the Workshop-Conference produce a volume on the round of Asian censuses, to be published under the auspices of the East-West Population Institute. The volume, which would bear the title Introductions to the 1970 Censuses in Asia, would summarize each country's census-taking procedures and problems. Everyone present approved of this suggestion. 35

50

51

52