4.1.1 Coverage and contents of population censuses
The population census is the most comprehensive population count. So far five modern censuses have been conducted in China in 1953, 1964, 1982, 1990 and 2000. The first three censuses were not taken on a regular basis and had very simple enquiries. It was not until 1986 that the Chinese government decided to organise regular decennial censuses after 1990 in the years ending in zero, supplemented by a one percent sample census carried out in the middle of the intercensal period (Sun 1997). Of the three most recent censuses, the 1982 census largely established the census-taking model for the following ones in census design, organisation, field operation and data processing. The later two included more census contents and used more advanced techniques in response to changing societies and rising demands from rapid socio-economic development.
China’s census counts people largely on a de jure basis, but with an increasingly number of de facto factors in more recent ones. The census instructions stipulate that all persons who hold the nationality of the People’s Republic should be enumerated at their permanent residence (changzhu di). This census-defined permanent residence is largely determined according to a person’s hukou-registered residential location, but is not necessarily the same, because the duration of time a person has lived away from the registered residence has also been taken into account. For example, in the 1982 and 1990 censuses, a person who was enumerated at his or her current residential location as permanent residence must have lived there with a hukou registered there, or have been resided there for more than one year with a hukou elsewhere, or have been away from the registered residence for more than a year. Otherwise, he or she should still be enumerated at the registered residence. In the 2000 census, the duration was reduced to six months. Meanwhile, the geographic boundary to determine the difference between a person’s current and registered residence was urban city or rural county both in the 1982 and 1990 censuses, but being changed to urban street, town and rural township in 2000.
The census unit is the household. The definition of two kinds of households is similar to that in the hukou system: the family household including both one in which many people live together with a family relation and the household of a single-person family; and the collective household. The respondents of the 10 percent long form of the 2000 census were also chosen based on households. There are two parts in the census questionnaire: one for the household, recording the information of each household member briefly, answered by the head of the household; the other for each member, recording individual information in detail, answered by each individual inside the family. The individual part of the census questionnaire also contains specific questions on childbearing information answered by only women of reproductive age.
The contents of the 1982 census were simple, collecting the basic demographic information, including name, relationship with the head of the household, sex, age, ethnicity, and current hukou status. But the census included socio-economic variables, such as education and employment-related items and used an internationally comparable occupational and industrial classification for the first time
(Yu 1997). The 1990 census made several important modifications of the one in 1982: one was in the definition of urban population. It proposed two different definitions to classify urban or rural population, one similar to the 1982 definition, the other including all people living within a city’s jurisdiction as urban population, regardless of their hukou registration status. The other important change was the addition of the items concerning migration, inquiring into individual residence five years ago and the reasons for migration (Wang 2000a). Both the 1982 and 1990 censuses used 1 July as the reference date.
There were more and new elements added to the 2000 census, as had been desired by scholars and policy-makers for a long time. For the first time, the census introduced both a short form and a long form in the enumeration. About 90 percent of households received the short from with only 19 items, including the questions answered by everyone in every household. The remaining 10 percent of households received the long form, which contained detailed questions on migration, housing, employment and childbearing, amounting to 49 items. Another important change was the definition of permanent residence, which used smaller geographic units and durations of stay to include people in their current dwelling, as mentioned above. In addition, the census organisers decided to shift the reference time to 1 November, which aimed to capture more rural-urban temporary migrants at their current residence. Also for the first time, the census introduced a new form to enumerate short-term temporary migrants (zhanzhu renkou) at their current dwelling, who left their permanent residence less than six months before and should still be enumerated there according to the census criteria.
All three most recent censuses collected fertility data from female respondents, obtaining information from women aged 15-64 in both the 1982 and 1990 censuses, and from women aged 15-50 in 2000. There are two childbearing-related questions: one about reproductive history, the number of children ever born and surviving; the other about the newborns in the year immediately before the census. Both the 1982 and 2000 census enquired into newborns in the preceding year, and the 1990 census gathered information on newborns in the preceding 18 months by three six-month periods. While all women of reproductive age answered the childbearing questions,
in the 2000 census only women in the long-form-sampled households were interviewed with these questions.
4.1.2 Data collection and quality control of population censuses
The SSB and its nationwide subordinates are responsible for the actual census design and implementation, but the efforts to mobilise the necessary economic and human resources to prepare and conduct censuses rely heavily on governments at all levels. Normally, one or two years before the census, the State Council establishes a special leadership group and the Population Census Office, composed of high-level officials from relevant governmental ministries and organisations at the central level. This structure is copied at each administrative level downward to the rural townships, urban towns and street communities to co-ordinate census activities at all levels. Rural villager committees and urban resident committees also create population census groups for practical performance (SSB 2001). Before each census, a variety of large-scale campaigns are organised to inform the general public about the census, and make necessary preparations, especially the hukou cleanup program. Millions of enumerators and supervisors are recruited and given the necessary training.
All censuses so far have used the field enumeration approach, and the enumerators, rather than respondents, fill in the census questionnaire. Even for the 2000 census, the organisers still considered a 12 percent illiteracy rate as a great problem for people filling the forms themselves (Zhang 2000). To count the huge population of more than one billion, the census mobilised 5-6 million census takers and supervisors in the field operation in the past three censuses. In fact, if the people who were actively involved in the publicity and various auxiliary activities were included, this number would be much larger, up to 10 million (Yu 1997). On average every enumerator interviews 200-300 persons. Most enumerators are temporarily recruited from government agencies, schools and factories, and had little or no statistical training. It is understandable that the degree of competence of the census takers and their efforts will greatly affect the census quality. But this could be compensated for by other favourable conditions in 1982 and 1990 censuses, but not in 2000.
There have been strict quality control standards, from preparation, field operation and cross-checking after household interviews, to data entry for the censuses. One important measure is to conduct an independent post-enumeration survey, as in the three most recent censuses (Wu 2001; Yang 1994). However, it appeared that these surveys are not very effective in revealing underreporting of young children. One possible explanation is that the post-enumeration survey emphasises the check of completeness of the whole population, paying less attention to underreporting of newborns and young children.