History of the Vietnam population censuses
Vietnam has a long tradition of censuses over several centuries. However, almost all these censuses, in reality, have only served to count the population to know how many people had to pay taxes, or to conscript soldiers to serve in wars. Thus, these counts were not regular and collected few details. During the period of
development of Socialism in the North and the struggle to unify the nation (1945-1975), the Democratic Republic of Vietnam implemented two population censuses covering the North, in March 1960 and April 1974. Right after the end of the war in 1975, at the beginning of 1976 the State organized a total population count for
all provinces in the South to serve National Assembly elections and reconstruction of the Nation.
The first real population census of the reunified Socialist Republic of Vietnam was implemented at the end of 1979. With the resources and technical capacity available at the time, the 1979 Census provided basic data of good quality to serve as the basis for the country’s development. The second national population census, regarded as the first modern Population and Housing Census in Vietnam, was implemented in April, 1989. This Census used internationally accepted concepts, design and data processingtechnologies. Many individuals and organizations that participated in the Census consider that the de jure population enumerated in the Census was nearly complete and the results were of high quality. In the following decade, this Census has provided a rich source of data on demography, society and the economy and has been used by many stakeholders.
The third National Population and Housing Census was implemented in April, 1999. Many features of the 1989 Census were included in the design and implementing guidelines for the 1999 Census. In addition, the 1999 Census added questions and expanded the contents to provide more comprehensive and detailed data.
The fourth National Population and Housing Census was implemented in April 2009. Compared to the previous censuses, the 2009 Population and Housing Census had a broader scope and a larger contents. The Communist Party and Government led and directed the Census and the people consented to and supported the Census. At the same time, this Population and Housing Census inherited and drew from the experience of the previous Censuses, especially the 1989 and 1999 Population and Housing Censuses. This was the third Census for which United Nations Population Fund (UNFPA) has provided technical and financial assistance for some important activities of the census.
The 2009 Population and Housing Census was implemented according to Prime Ministerial Decision No. 94/2008/QD-TTg dated 10 July, 2008. This was the fourth population census and the third housing census implemented in Vietnam since the nation was reunified in 1975. The Census aimed to collect basic data on the population and housing for the entire territory of the Socialist Republic of Vietnam, to provide data for research and analysis of population and housing developments nationally and for each locality. It responded to information needs for assessing implementation of socio-economic development plans covering the period 2001 to 2010, for developing the socio-economic development plans for 2011 to 2020 and for monitoring performance on Millennium Development Goals of the United Nations to which the Vietnamese Government is committed.
Kind of Data
Census/enumeration data [cen]
Unit of Analysis
The survey contents appeared in two types of survey questionnaire: the complete census form (short-form) and the census sample survey form (long form).
Contents of the complete census form
a) On population:
- Individual information (full name, sex, month and year of birth/age);
- Relationship to household head;
- Current school attendance;
- Educational attainment;
- Ethnicity and religion;
- Literacy status.
b) On housing:
- Current housing situation;
- Structure of dwelling and primary construction materials;
- Area of dwelling;
- Year the dwelling was put into use.
Besides information collected in the complete census form, the census sample survey form collected additional information as follows:
a) On population:
- Place of usual residence 5 years before the Census;
- Disability status;
- Marital status;
- Highest professional/technical qualifications;
- Economic activity status in the 7 days prior to the Census.
b) On fertility for women aged 15-49:
- Whether or not the woman ever gave birth;
- Number of children born, number who survived and number who died;
- Month and year of birth, number of boys, number of girls for the most recent birth.
c) On deaths:
- Mortality situation in the household;
- Individual information about the death;
- Cause of death and maternal mortality.
d) On housing:
- Use of dwelling;
- Ownership of dwelling;
- Primary fuel source for lighting and cooking;
- Main water source for drinking and eating;
- Type of toilet used;
- Some basic amenities of the household.
The 2009 Population and Housing Census enumerated all Vietnamese regularly residing in the territory of the Socialist Republic of Vietnam at the reference point of 0:00 on 01 April, 2009; Vietnamese citizens given permission by the authorities to travel overseas and still within the authorized period; deaths (members of the household) that occurred between the first day of the Lunar Year of the Rat (07 February, 2008) to 31 March, 2009; and residential housing of the population.
Population and housing censuses were implemented simultaneously taking the household as the survey unit. The household could include one individual who eats and resides alone or a group of individuals who eat and reside together. For household with 2 persons and over, its members may or may not share a common
budget; or be related by blood or not; or marital or adoptive relationship or not; or in combination of both. The household head was the main respondent. For information of which the head of household was unaware, the enumerator was required to directly interview the survey subject. For information on labour and employment, the enumerator was required to directly interview all respondents aged 15 and older; for questions on births, the enumerator was required to directly interview women in childbearing ages (from 15 to 49 years of age) to determine the responses. For information on housing, the enumerator was required to directly survey the household head and/or combine this with direct observation to determine the information to record in the forms.
Producers and sponsors
General Statistics Office
Ministry of Planning and Investment
United Nations Population Fund
Technical and Financial Assistance
In the 2009 Population and Housing Census, besides a full enumeration, some indicators were collected in a sample survey. The census sample survey was designed to: (1) expand survey contents; (2) improve survey quality, especially for sensitive and complicated questions; and (3) save on survey costs. To improve the
efficiency and reliability of the census sample data, the sample size was 15% of the total population of the country. The sample of the census is a single-stage cluster sample design with stratification and systematic sample selection. Sample selection is implemented in two steps: Step 1, select the strata to determine the sample size for each district. Step 2, independently and systematically select from the sample frame of enumeration areas in each district to determine the specific enumeration areas in the sample.
The sample size of the two census sample surveys in 1989 and 1999 was 5% and 3% respectively, only representative at the provincial level; sample survey indicators covered fertility history of women aged 15-49 years and deaths in the household in the previous 12 months. In the 2009 Census, besides the above two
indicators, many other indicators were also included in the census sample survey. The census sample survey provides data representative at the district level. When determining sample size and allocation, the frequency of events was taken into account for various indicators including birth and deaths in the 12 months prior to the survey, and the number of people unemployed in urban areas, etc.; efforts were also made to ensure the ability to compare results between districts within the same province/municipality and between provinces/ municipalities.
Stratification and sample allocation across strata
To ensure representativeness of the sample for each district throughout the
country and because the population size is not uniform across districts or provinces, the Central Steering Committee decided to allocate the sample directly to 682 out of 684 districts (excluding 2 island districts) throughout the country in 2 steps:
Step 1: Determine the sampling rate f(r) for 3 regions including:
- Region 1: including 132 urban districts;
- Region 2: including 294 delta and coastal rural districts;
- Region 3: including 256 mountainous and island districts.
Step 2: Allocate the sample across districts in each region based on the sampling rates for each region as determined in Step 1 using the inverse sampling allocation method. Through applying to this allocation method, the number of sampling units in each small district is increased adequately to ensure representativeness.
The formula used to calculate the sample rate for each district in each region is provided on page 22 of the Census Report (Part1) provided as external resources.
Sampling unit and method
The sampling unit is the enumeration area that was ascertained in the step to delimit enumeration areas. The sampling frame is the list of all enumeration areas that was made following the order of the list of administrative units at the commune level within each district. In this way, the whole country has 682 sample
frames (682 strata).
The provincial steering committee was responsible for selecting sample enumeration areas using systematic random sampling as follows:
Step 1: Take the total of all enumeration areas in the district, divide by the number of enumeration areas needed in the sampleto determine the skip (k), which is calculated with precision up to 1 decimal point.
Step 2: Select the first enumeration area (b, with b = k), corresponding to the first enumeration area to be selected. Each successive enumeration area to be selected will correspond to the order number: bi = b + i x k ; here i = 1, 2, 3…. Stopping when the number of enumeration areas needed has been selected.
General weights were calculated based on probability/weights. Pages 23-27 of the Census Report (Part 1) provided as external resources detail the weight calculation procedures.
Dates of Data Collection
Data Collection Mode
The team leader was the workforce that directly managed the work of enumerators. For recruitment of team leaders priority was placed on the following criteria: (1) ability to organize team work, and (2) commitment to allocate 100% of their time to managing the enumerators during the implementation of the survey in their locality. Besides the team leader, there was a workforce of supervisors at all levels from central, provincial, to district levels with primary responsibility for supervising the activities of the network of team leaders and enumerators, monitoring and providing technical support for the lower level Steering Committees.
Data Collection Notes
The census questionnaires and materials were tested through three rounds of pre-testing and one pilot test prior to being officially launched for the Census. Three pre-tests were implemented to refine the survey protocol, census questionnaires and other technical guidelines, with pre-test locations at the 6 provinces/cities under directly to the Central Government including Ha Noi, Vinh Phuc, Thai Nguyen, Quang Tri, Ho Chi Minh City and Ben Tre.
In August and September, 2008, the final Census pilot was successfully implemented in 48 enumeration areas belonging to 4 provinces (12 enumeration areas per province) representing all regions including: Yen Bai representing the Northern Midlands and Mountains; Nam Dinh representing the Red River Delta; Quang Binh representing the North and South Central Coast and Central Highlands; and Tien Giang representing the Southeast and Mekong Delta. Provinces and municipalities and relevant Ministries and sectoral agencies contributed staff to participate in the pre-testing and final pilot test. The results of these surveys, along with experience gained, served as the basis for refining the survey protocol, census questionnaires, census procedures and technical guiding documents for collection, processing and synthesizing results of the Census.
Recruitment and training of enumerators
Training aimed at providing the Census with knowledgeable personnel able to direct and train operational and field staff at all levels, all the way down to the team leader and enumerator level. Pages 13-15 of Part 1 of the report provided as external resources details the training process for census implementation.
In order to be recruited, enumerators had to meet criteria such as a strong sense of responsibility, ability to write numbers and letters legibly, with educational attainment of grade 10/12 or higher (in highland and remote areas, the lowest acceptable educational attainment was grade 7/12) and preferred ages from 25 to 50 years. Enumerators could be farmers, civil servants, students, teachers, military border control staff, etc. Almost all interviewers were recruited from the local area. In a few areas in the high mountains, or extremely remote areas where the majority of inhabitants were ethnic minorities with limitations in educational attainment, the enumerators were brought in from other areas. In these areas, the local people (usually hamlet leaders) helped to guide enumerators around and interpret for them.
The questionnaires and survey materials were designed and tested three times before final approval.
The 2009 Population and Housing Census applied Intelligent Character Recognition technology/scanning technology for direct data entry from census forms to the computer to replace the traditional keyboard data entry that is commonly used in Vietnam at present. This is an advanced technology, and the first time it had been
applied in a statistical survey in Vietnam. Preparatory work had to be done carefully and meticulously. Through organization of many workshops and 7 pilot applications with technical and financial assistance from the UNFPA, the new technology was mastered, and the Census Steering Committee Standing Committee approved use of this technology to process the entire results of the 2009 Population and Housing Census. The Government decided to allocate funds through the project on Modernization of the General Statistics Office using World Bank Loan funds to procure the scanning system equipment, software and technical assistance.
The successful use of this technology will create a precedent for continued use of scanning technology in other statistical surveys
After checking and coding at the Provincial/municipal steering committee office, (both the complete census and the census sample survey), forms were checked and accepted then transferred for processing to one of three Statistical Computing Centres in Hanoi, Ho Chi Minh City and Da Nang. Data processing was implemented in only a few locations, following standard procedures and a fixed timeline. The steering committee at each level and processing centres fully implemented their assigned responsibilities, especially the checking, transmitting and maintenance of survey forms in good condition. The Central Steering Committee collaborated with the Statistical Computer Centres to set up a plan for processing and compiling results, setting up tabulation plans, interpreting and synthesizing output tables, and developing options for extrapolating from sample to population estimates.
The General Statistics Office completed the work of developing software applications and training using ReadSoft software (the one used in pilot testing), organized training on network management and training on systems and programs for logic checks and data editing, developed a data processing protocol, integrated
these systems and completed data flow management programs. The General Statistics Office collaborated with the contractor, FPT, to develop software applications, train staff, testl the system and complete the programs using the new TIS and E-form software.
Compilation of results was implemented in 2 stages. In stage 1 data were compiled from the Census Sample Survey by the end of October, 2009, and in stage 2, data were compiled from the completed census forms, with work finalized in May 2010.
Estimates of Sampling Error
Estimates from the Census sample survey were affected by two types of error: (1) non-sampling error, and (2) sampling error. Non-sampling error is the result of errors in implementation of data collection and processing such as visiting the wrong dwelling, interviewing the wrong household, mis-understanding of questions by respondents, other errors on the part of respondents or enumerators and data entry. Although many efforts were made to minimize these errors while implementing the survey, non-sampling errors cannot be completely avoided are difficult to evaluate statistically.
On the other hand, sampling error can be evaluated statistically. The sample of respondents in the Census sample survey is only one of many possible samples that could be selected from the total population using the same sample design method and required sample size. Each of these possible samples could give different results from the sample actually selected. Sampling error results from variation in results from the many different possible samples. Although it is not possible to know this variation precisely, it can be estimated from the sample survey results.
Sampling error is usually measured using standard errors for specific statistical indicators (means, proportions etc.). The standard error is, in fact, the square root of the variance. Standard errors can be used to calculate confidence intervals which contain the true value for the population. For example, for a given statistical indicator calculated from the sample survey, the true statistical value will fall within the interval plus or minus two times the standard error for that indicator with confidence equal to 95% for all possible samples of the same size and sample design.
If the sampling unit is selected following simple random sampling, then it is possible to use formulas to directly calculate standard errors for the sample. However, the Census sample survey was designed with strata, and therefore a more complicated formula must be used. Computer software that could be used to calculate standard errors for stratified samples includes the standard error calculation module of ISSA or STATA. These programs use the Taylor linear expansion method to estimate variance for estimates of means and proportions for sample surveys (See the formula on page 28 of the Census Report (Part1) provided as external resources).
Sampling error in the Census sample survey is calculated for a few selected key indicators. Results are presented in an appendix for estimates at the national, urban and rural, and 6 socio-economic region levels and for 63 provinces/municipalities. For each variable, the statistical estimate (R), standard error (SE), relative standard error (SE/R) and 95% confidence interval (R±2SE) are presented in Annex 4 (provided as external resources).
The confidence interval (for example, when estimating the indicator sex ratio at birth) can be explained as follows: the national estimate for sex ratio at birth is 110,5 boys to 100 girls with a standard error equal to 0.54. Therefore, for a 95% confidence interval, adding and subtracting 2 times the standard error to the
estimate gives us, 110.5±2×0.54. With a high probability (95%) the national sex ratio at birth lies in the interval from 109.5 to 111.6 boys per 100 girls.
One extremely important requirement of the 2009 Population and Housing Census was that the quality of the survey data be assured. Implementation of the 2009 Population and Housing Census reflected results of human resource development at the General Statistics Office over many years. Much of the design
and implementation work during the 2009 Population and Housing Census benefitted from and involved further development of methods applied in the 1989 and 1999 Population and Housing Censuses, from the census publicity campaign through the training program and survey regulations. The results are that the General Statistics Office has gained a high level of self-reliance in its ability to organize the 2009 Population and Housing Census with reduced financial and technical assistance from international and foreign organizations. Right after completing the data collection, the Central Steering Committee Standing Committee implemented many methods to assess and verify outcomes of the Census and the results are as follows:
• Verification and supplementary enumeration through the hotline:
The Central Steering Committee Standing Committee collaborated effectively with Vietnam Television and Voice of Vietnam radio to announce to the people throughout the country that they should help to provide information to identify all survey subjects who may have been missed so supplementary enumeration could be performed. By 20 April, 2009, the Central Population and Housing Census Steering Committee Office had received feedback from 975 households throughout the country who called the hotline. The results were checked, and 744 households were interviewed to add to the census results, while the remaining 231 households were found to have already been enumerated through other household members or through indirect methods. Although the number of households that were supplemented was not large compared to the nation’s 22 million households, through this activity, the households and Census Steering Committee at all levels showed their consent and trust in the Census results.
• Post-enumeration survey:
Right after the survey work was completed, a sample post-enumeration survey was undertaken in order to evaluate the accuracy of the Census data. A systematic sample involving 60 enumeration areas was randomly selected for reinterview organized directly by the Central Steering Committee Office.
With the objective of evaluating the completeness of the Census, all usual residents of each household were asked four questions : (1) full name, (2) relationship to household head, (3) sex, (4) month and year of birth or age. The responses to these questions were compared to the survey forms (after they were
transferred to the higher levels) to find any cases that were missing from the Census or in the post-enumeration survey. Results of the post-enumeration survey indicated a net error rate of -0.3% (equal to the difference between the undercount rate and the duplication rate), or equivalent to about 258 000 people (see detailed table on page 18 of Part 1 of the Census Report - provided as external resources). If we compare this rate with the error rate of 1.5% to 4.5% of the total population normally found in population and housing censuses in other countries, and at the same time comparing to the previous Census in Vietnam, completeness of this Census is assessed as high.
• Evaluation of the quality of recording in forms:
Immediately after the enumeration was completed, the Central Steering Committee Office collaborated with the Statistical Computing Centre to organize teams to check the quality of recording in the census forms. The results of this exercise indicate that the quality of recording in forms met requirements, and
ensured the ability to use scanning technology for processing.
• Checking focal sites for subjects with a high likelihood of being omitted:
In order to have a stronger basis for releasing preliminary results of the Census, in the first half of July, 2009, the Central Steering Committee Standing Committee decided to undertake sentinel checking for 3 types of subjects with a high probability of being missed. These subjects included: (1) students living in dormitories, (2) students living in boarding houses in the enumeration area, and (3) workers working in industrial zones who rent housing from the local population or who live in tents or other temporary housing of the project or production enterprise far from residential centres. Particularly for Hanoi, Ho Chi Minh City and Binh Duong, places with a large number of in-migrants, the Central Steering Committee directly implemented the checking.
Results of checking indicate that students in dormitories were rarely omitted, because each dormitory has a management board that knows clearly the situation of their housing and information on the individual students, and members of the management board are usually the team leaders or enumerators involved in the survey work. For the remaining two groups, the checking revealed that there is little duplication or omission of these groups, i.e. the undercount and duplication rates are about the same as those found in the post-enumeration survey for the general population.
Besides the methods for checking and post-enumeration mentioned above, during the process of undertaking the Census, the Central Steering Committee organized checking to review the results of drawing maps and listing dwellings 2 times, and to review the preliminary compilation of results 3 times. Thus, through many different methods of checking and post-enumeration surveys, we come to the assessment that the 2009 Population and Housing Census has collected reliable and stable results, appropriate with the effort and work contributed by the political system, the Steering Committees at all levels, the various sectors and the active
participation of the people throughout the country.
General Statistics Office
Ministry of Planning and Investment
General Statistics Office
Ministry of Planning and Investment
World Bank Microdata Library
Use of the dataset must be acknowledged using a citation which would include:
- the Identification of the Primary Investigator
- the title of the survey (including country, acronym and year of implementation)
- the survey reference number
- the source and date of download
Disclaimer and copyrights
The user of the data acknowledges that the original collector of the data, the authorized distributor of the data, and the relevant funding agency bear no responsibility for use of the data or for interpretations or inferences based upon such uses.