Chinese(GB)
China Statistical Database
E-Mail
Luigi Biggeri
Senior Project Advisor, China/Italy/FAO Projects,
University of Florence, Italy
Abstract: The First Agricultural Census conducted in China in 1997 is the world’s largest and most complex statistical survey ever undertaken. The census was a significant and exciting challenge for applied statisticians and agricultural economists. At its conclusion, it is important to review the work done, and the choices made in methodology for its implementation in order to pinpoint the reasons for its success. Several issues should be examined, which could provide lessons for the future of agricultural census-taking, and the improvement in agricultural statistics for China and other countries.
Keywords: China, Agricultural Census, Planning and Research Strategy, Data Collection.
1. Introduction
Preparation for the First Agricultural Census in China began in 1983, in response to a request for technical assistance by the Chinese Government to the Food and Agriculture Organization of the United Nations (FAO). In 1987, FAO, Italy and China agreed to a four-year project, having the long-term objective of providing China with the technical assistance to carry out a National Agricultural Census. Continuous support was ensured within this framework throughout the implementation of three other projects, the realization of the Chinese Census and up to the present.
As Senior Project Advisor for the projects, I had the great opportunity of sharing 13 years of exciting and challenging experience with Chinese colleagues and others. It has been of extraordinary benefit for me professionally and individually. I would like to thank FAO, Italy, the Government of China and all the Chinese and FAO officers and international experts for imparting their expertise to me, and for the work done together. The Chinese Government and my colleagues are to be heartily congratulated now that the first Chinese Agricultural Census has been completed, and is considered a success by statistical agencies and researchers.
It is important to review the work done and the methodological choices made for the implementation of the census, in order to pinpoint the reasons for its success and examine certain issues[1][1]. This should provide lessons for future agricultural census-taking and the improvement of agricultural statistics in China and in other countries.
This paper is structured as follows: First, the features and content of the planning and research activities carried out for the definition of the census design and programme are presented as the most critical factors in ensuring the success of a census. Particular reference is made to the pilot censuses implemented, and to the involvement in the discussions, both by producers and users of census data at the national and international level. This will highlight the weaknesses and strengths of the work carried out. Second, the paper focuses on some of the main aspects and activities of the census (census scope, concepts, definitions and data collection), to point out particular issues and questionable items (list of the units, coverage errors and agricultural land use estimation), to clarify the choices made, and the quality, intent and correct interpretation of the data collected. Suggestions for the implementation of the next agricultural census and current statistics in China will be made. Finally, the paper concludes that analyses and research on the validity and quality of the census data, as compared to current statistics, must be continued and enhanced by the substantial involvement of the users. The data dissemination strategy should help accomplish this.
2. The “Scientific” Framework and Process to Define the Final Census Design and Programme
2.1 The Need For a Systematic Research-evaluation Approach
The methodology followed to organize and conduct the Chinese Agricultural Census, and the main factors of the successful experience of the census (learning from international experiences, conducting pilot censuses, training a large number of statisticians and interviewers, strong and powerful organization, wide range of publicity, providing adequate hardware and software for data processing, etc.) have been explained by Zhu Xiangdong (1995 and 1998), and will be illustrated during the International Seminar.
Planning and research strategy, and activities carried out in defining the final census design and programme was essential to ensure the success of the census. The implementation of a complex statistical survey, in particular, of an agricultural census, must be viewed as the implementation of an information system to capture relevant information on the actual situation, maximizing data utility and information needs[2][2].
In order to obtain the established objectives, the preparation of a census design and programme has therefore to be considered as a continuing scientific experiment,. Following a planning strategy and starting from user needs, it is essential to do research, make data collection experiments, analysis, verification, control and feedback; and to repeat this process-cycle up to the definition of a census design and programme, which will capture reality as fully as possible. Therefore, in order to choose between the different conceptual and operational steps and decisions, and to suggest which of these seems, case by case, the most appropriate, a systematic evaluation approach as an iterative process of study, data collection, research function (checks), verification of user needs satisfaction, and feedback must be implemented (see fig. 1).
Figure 1:. A simplified production process for the definition of the census design and programme and their evaluation
2.2 The Systematic Scientific Evaluation Process Followed for the Design of the Agricultural Census
In the preparation of the final design and programme for the Agricultural Census in China, the above-mentioned general scientific strategy has been followed consistently, as illustrated by the description of the different steps implemented to this end in the following scheme:
2.2.1 Scheme 1: Essential phases in the definition of the Agricultural Census design and programme, and their evaluation:
1. 1. First phase of study, training and research (1987-1990).
· · Utilizing the international documents (FAO, 1986), and taking into account the situation in China, a study unit was implemented to prepare a first draft of the census designs and programmes of other countries.
· · The Food and Agricultural Statistics Centre (FASC), was established in Beijing to create training and research capabilities in preparation for the Agricultural Census.
· · A six-week workshop and five six-month training courses for 185 national and provincial statisticians was carried out.
· · An integrated in-country and abroad system of training, (fellowships and study tours) was implemented during all the different FAO/Italy/China projects.
· · An audio-visual unit for the production of educational audio-visual cassettes was established to ensure teaching homogeneity for all the planned training courses.
· · Preparation was started for the first draft programme for the census.
2. 2. First pilot census (March 1991).
· · A pilot census in Weixian County, Hebei Province, was conducted, using complete enumeration (about 101 000 holdings), to implement a field experiment of all the activities of an agricultural census, in order to verify the design, definitions, methods, etc., and to acquire practical experience on the execution of the census. Only two questionnaires were used, one for holdings, and one for land resources and land utilized at the village level, (the intention was to undertake the agricultural census in 1994/95).
3. 3. National summing-up discussions and international expert consultation on agricultural censuses (September-October 1991): feedback.
· · Provincial and national summing-up discussions were held, and one international seminar on pilot census results was organized to disseminate the results of the pilot census, discuss the findings, solicit studies and research, involve as many statisticians and other researchers as possible (to better inform users) and propose modifications of the first draft design of the census. Various research papers were presented at the different events (Zhu Xiangndong, 1992; Shao Zongming, 1992). (It was decided to postpone the taking of the agricultural census).
4. 4. Phase of expansion of training, study and research (1991-1994).
· · The census design and programme was revised, in particular the content, definitions and questionnaires. The training and research activity at the territorial level was diffused to cover the entire country. A new set of field experiments was prepared.
· · Regional Food and Agricultural Statistics Centres (RFASCs) were set up to expand the methodological and technical ability of implementing the agricultural census nationwide (starting with six RFASCs. Before the census began, 22 RFASCs and nine provincial data centres were established).
· · A set of training courses were organized in agricultural statistics, agricultural census techniques, data collection and processing and statistical and agricultural economic analysis. This included field demonstrations, and emphasis on problem-solving skills to improve the quality and ability of the rural statistical personnel. The large-scale training programme was organized in a top-down pyramid programme to create a multiplier training effect all over China. Some 4 000 agricultural statisticians and 8 000 statistical agents were trained to be used as cadres for the organization of the census and agricultural surveys.
· · Training material prepared by international experts, FAO staff members and others, books on the topics of agricultural census, data processing, agricultural economics, etc., were translated into Chinese and disseminated to the RFASCs. Chinese technicians prepared the appropriate manual and books, videotaped cassettes, etc. to facilitate teaching, and to stimulate research in the field.
· · Research was conducted to find better methodological and technical solutions for data processing, data analysis and complex data tabulations. (In January 1994, the State Council decided to conduct the First National Agricultural Census in China in January 1997).
· · A programme for the organization and implementation of a set of pilot censuses was prepared.
· · An expert consultation meeting on the programme for the organization of an agricultural pilot census was taken in Rome (20-21 October 1994), to discuss all the technical problems, in particular: (i) the use of different questionnaires to obtain information on different units (household holdings, non-household holdings, non agricultural household and towns and townships enterprises), (ii) the use of complete or sampling enumeration and (iii) the system of data quality control.
5 Sixteen pilot censuses conducted in different provinces (January-March 1995).
· · Pilot agricultural census were conducted, as a quasi-experimental design, by the RFASCs in 16 counties to observe the differences in agro-climatic and agricultural characteristics, socio-economic organization and the level of rural development, as well as to test various methodological and technical alternatives. The pilot censuses, which covered about 148 000 units, were an opportunity to gain experiences, identify problems and test the feasibility of collecting the information necessary to meet the needs of principal users in China, as well as those of the international community. Some preliminary analytical exercises on pilot census data were planned and carried out to verify how appropriate they were in answering relevant policy questions. Thus, pilot census activities are to be considered as both a training and a methodological activity. Changes in the census design and data collection methods were proposed. A summary of the organization and characteristics of the pilot censuses is reported in Annex 1.
6 6 International Seminar on “Taking the First Agricultural Census in China: Aims, Problems, Methodologies, Challenges, Opportunities” (13-17 November 1995).
· · The seminar was an important opportunity to discuss at the international level the results, findings and recommendations of the pilot censuses, to exchange experiences on the most appropriate methodologies used in other countries and to discuss data users’ points of view. Research carried out during the pilot census on the data collected, and papers were presented (FASC, 1995).
7 7 Two pilot censuses conducted in Guangdong Province and in Beijing, and a national pilot census conducted in each province (April 1996).
· · The two pilot censuses covered two townships in Guangdong (enumerating 14 000 agricultural holdings) and 23 towns and townships in Beijing (covering 21 500 agricultural holdings). Other field tests on specific topics were conducted by small pilot censuses in each province before the beginning of the census, in particular, on questionnaires and data collection methods (filling out, etc.).
8 8 Final census design and programme (July 1996).
· · The final design and programme for taking the agricultural census was prepared. The implementation of the census entered its most important phase, with the adoption of updated technology, having data quality as the priority. The starting date of the First Chinese Agricultural Census in China was the first of January 1997.
9 9 External controls and evaluation analysis during and after the census field work (January-April, 1997).
· · To verify the validity of the implementation of the census programme, two important activities were planned and carried out:
· · An “International Census Committee” was established to evaluate the fieldwork implementation of the census, how the census met world standards and to give advice on the analysis of census data. A mission to China from an international committee composed of 11 experts was organized from 12 to 22 of January 1997, to visit the census work-areas in different provinces.
· · A Post Enumeration Survey (PES) was implemented in March-April 1997, which included visits to the fieldwork areas by international consultants. The PES took into account the suggestions of the above-mentioned international committee.
10. 10. Dissemination of census results, and of detailed data to encourage research analysis on the quality and economic significance of data collected (1998-2000).
· · Using the Optical Character Recognition (OCR) technology, the census data entry and verifications were finished by January 1998. Therefore, the census results were promptly available. In February 1998, the first results were disseminated but lacked specific data because of the need for their reconciliation with current statistics (in particular, land-use and livestock inventory data). In November 1999, all of the census results were published and disseminated throughout China, as illustrated in detail in other papers presented at the seminar. The main dissemination activities were FASC, the National Bureau of Statistics of China (NBS) and the Provincial Statistical Bureau (PBS). Other institutions carried out analyses, and issued publications at the provincial and national level, as well as organized workshops/seminars.
· · Award prizes for the best Chinese papers on agricultural census data analysis were established. One hundred forty-nine Chinese papers were prepared and proposed for prizes.
· · In order to disseminate the census results at the international level, the following activities were implemented:
a) a) Two events (press conferences) were organized at FAO in Rome (26 February 1998) and in Beijing (4 December 1998).
An “Abstract”, written in English, containing 240 tables was prepared, including maps utilizing the Geographical Information System (GIS) technology and software, copies of the questionnaires and a glossary of the concepts and definitions used. It was disseminated in hard copy by CD-ROM (National Agricultural Census Office (NACO) and FASC, 1999) and on the Internet NBS Web page, where five communiqués were also posted. In addition, 120 supplementary tables having more detailed census results, as well as a report on the methodology used are now available. These will be included in the CD-ROM, and on the NBS Web site.
b) b) A one percent sample data base of the individual records of the census questionnaire A601 has been prepared, without risk of statistical disclosure, to be used by national and international experts for the analysis of data quality and census results from an agricultural economic point of view (it is not possible to use the full database, which is about 210 GB). The 1 percent sample database on the NBS Web site is to be accessed by all interested users, with a query facility through which tabulations and statistical computation can be requested.
· · The International Seminar on Chinese Agricultural Census Results to be held in Beijing from 19 to 22 September 2000 (where this paper was presented).
2.3 Issues and Lessons
As is clear from the scheme presented, the iteration of the study cycle to design the census was repeated three times. Moreover, three subsequent drafts and the final census programmes were prepared (1990, 1992, 1994 and 1996). As a consequence, the Agricultural Census of China has been tailored to its unique situation.
Rich improvements of the programmes were obtained through the implementation of the systematic “scientific” process design, which show the extreme importance of the in-depth study of the Chinese agricultural and rural organization, the extensive training activity throughout China, and the involvement of as much staff as possible at the national and territorial levels in order to study and research the data collected by the pilot censuses. Moreover, there were extensive discussions of the methodological and technical problems both at national and international levels.
Two points that slightly affected the process design should be mentioned: (i) the insufficient knowledge of the specific organizational framework and management of the agricultural and rural economic units in the different provinces of China, and (ii) the lack of reliable lists of agricultural holdings (a situation recurrent in other countries), which conditioned all the programme discussions and design.
The study of process design takes years of planning and organization, and the lessons learned are: (a) to study the necessary changes in the programme well in advance; (b) to organize specific training activities to update the knowledge and ability of staff and researchers; (c) to follow changes in the organizational framework, and management of the rural and agricultural sector by appropriate sample surveys on the structure of the holdings.
3. Concepts, Definitions and Data Collection: Main Options and Issues
The International Committee, which was established to evaluate the fieldwork of the census, and how the census met prevalent world standards, affirmed that the work done by the Chinese in conducting the Agricultural Census was carried out satisfactorily and diligently. Data quality was consistently high in all the provinces of China. There were numerous successful choices and activities and factors which led to the positive results of the census (Zhu Xiangdong, 1998).
The following focuses on the main aspects and activities of the census - census scope, concepts, definitions and data collection - to point out several debatable issues, and clarify the reasons for choices made, the quality of the data collected, their significance and correct interpretation.
3.1 Census Scope and List of the Units
3.1.1 Census scope and definition of the units. The scope of the Chinese Agricultural Census was larger than that of agricultural censuses in other countries. The integration of agricultural activities with other economic activities in the rural sector in China was, and is, of great interest and importance in defining the rural situation. Besides the data on the structure of the agricultural sector, the census collected information on various other aspects of the rural economy, using different questionnaires (NACOC and FASC, 1999).
The integration of the holdings having enterprises engaged in other economic activities is relevant to the Chinese situation. For almost ten years, China has undergone an economic transition from a collective to a “socialist market economy”. This is also true of the rural sector. The agricultural and socio-economic reform policies introduced in China, with the diffusion of the “responsibility system” of farm households, and the development of rural industries, has made relations between the agricultural and rural sector more complex (Razelle S., 1994; Biggeri M., 1998). The decision-making structure in the Chinese rural economy shows that production decisions in agricultural farms and rural industries are a result of cooperation between the local village leaders who manage the industries, and farm households. The latter allocate labour force and other inputs to the agricultural and industrial sector of the village. Therefore, the reforms have created a complex environment in which village leaders and farm households must work together in order to maximize their common welfare. This situation, to my knowledge, is unique.
Chinese national and local governments clearly needed the detailed information on these relationships concerning smaller areas, which was obtained by the census and subsequent surveys. This made the Agricultural Census more complex. Six different questionnaires were required for the different units (actors), as was verified during the pilot censuses and affirmed during the International Seminar held in 1995 (FASC, 1995). Detailed data was needed on village and township involvement in agriculture, on the availability of infrastructure and environmental data, as well as information on community “facilities”. Consequently, the census had to collect more items, which meant more work for the census takers. The census thus became a kind of “rural census”.
3.1.2 List of the units and possible coverage errors. Since the adoption of the reform, the agricultural production units went from fewer than six million production teams to about 230 million holdings. At the same time, the form of agricultural management and production scale changed, making it impossible to collect a complete set of statistical data from such a large number of reporting units using current statistical operations. The existing comprehensive reporting system had lost its base after the rural economic reform, therefore the reported information lacked reliable sources, which seriously affected its accuracy, and the lists of the reporting units (holdings) could not be used for the census.
The issue was solved during the taking of the census. As usual, a pre-enumeration survey was carried out, both for the field enumeration, and as a basis for data quality control.
It is possible that coverage errors (Onate B. T., 1995) occurred in the China Agricultural Census, because, among other difficulties, in the Chinese agriculture sector, there are: (i) very small areas of farmland; (ii) high fragmentation of the farms and farmland; (iii) a frequency of mixed and associated crops and (iv) a frequency of successive crops harvested during the year. Nonetheless, the exhaustive instructions established to assure consistency of the data collected, to avoid duplications, overlapping and over-counts or under-counts, seem to have resulted in few errors and non-responses, according to the results of the Post Enumeration Survey (PES) (see the paper on the PES presented at the seminar).
3.2 Census Organization
The organization at all levels for the development and implementation of the Agricultural Census in China was impressive, and was certainly one of the main reasons for its success. The basis for this was: strong guidance, importance of guidance groups at the various levels, administrative power, enthusiasm, the large number of personnel available for organizational matters and the strong political support, which was extremely valuable in obtaining the necessary financial resources, as well as providing an unified promotion of the census at all levels.
In China, there is a complete set of administrative systems from the county level down to the township and village level. This is a powerful facility for the collection of census data but sometimes provoked difficult situations. For example, the following concerns arose: (i) at the provincial and lower levels, the local authorities sometimes requested too many local items be added to the questionnaires that were not always justified by the local situation; (ii) the presence of the village leaders during the interviews in order to secure cooperation may have affected (or dictated) the responses given by farmers, who may have only wanted to let the local leaders know certain things about their operations (this procedure may have been regarded by most farmers as not sufficiently separated from the existing bureaucratic process, despite reassurances given in the previous promotion); (iii) too much administrative data utilized to adjust households reports, with possible consequent biases.
From this point of view, it is evident that the collection of data during the agricultural census is still, in part, somewhere between a statistical survey and an administrative registration, which was done necessarily in this manner because of the educational level of the rural population, and their insufficient “trust” in statistical surveys. (Before the census, the State Council stated that the census results would not be used as a basis to changing current rural and agricultural economic policies).
3.3 Data Collection Issues and Possible Over or Underestimation of Certain Key Variables
When the agricultural census results were disseminated for the first time at the press conferences in 1998, there were several apparent differences between these and the data from current statistics, which is usually the case in other countries (Vogel F. A., 1999). These differences referred, in particular, to the land use and livestock inventory data, and were quite diversified at the provincial and local levels.
It is important to remember that the previous current statistics that referred to agriculture holdings and agricultural land were to be considered inaccurate (as was clearly recognized by the State Statistics Bureau (SSB) in its Statistical Yearbook, 1993, p. 4).Therefore, the statistical staff were unable to directly collect data from agricultural holdings, and the complete reporting information was based on estimation at the village and township level. Several of the figures are related to the performance of the local authorities, e.g. output, family planning items, sown area, etc. and others are related to state taxation, e.g. cultivated land and agricultural taxation. Moreover, it was difficult to avoid local authorities or persons focusing on their own regional or departmental interests, thereby influencing the collection and estimation of data.
Reasons for the technical differences between the census data from different questionnaires, and between the census data and the current statistical data are reported in the “Explanatory Notes on Main Items, and in the “Guidelines and Methodological Considerations” included in the Annex of the Abstract (NACOC and FASC, 1999). They are also stated in the Technical Report on the census that is currently being published.
Moreover, reasons for possible errors in the census data, with possible underestimation, were mentioned by the International Committee by Chinese colleagues who found that, in some areas: i), households had difficulty replying to sensitive items (cultivated land and number of livestock) because of taxes to be paid; ii) enumerators were not meticulous and iii) there were transportation problems. However, none of these issues were considered to be serious deterrents to the general quality of the data collected.
Chinese colleagues were aware of these problems, and of the difficulty in obtaining reliable data from the household interviews. As a result, various experiments were carried out during the pilot censuses (FASC, 1995).
Focusing on the agricultural land area, the Chinese considered the current statistics to be underestimated. Thus, during the pilot censuses, it was attempted to collect data directly from the household holdings from the questionnaires that referred to those units. However, as was found during the pilot censuses, it was impossible to obtain reliable estimation of the total cultivated land through the holding registrations or farmers’ replies.
For example, in Dongto County in 1995, cultivated land, as estimated by the pilot census was: = 702 000 mu; current statistics = 647 000 mu; and Department of Land Management = 968 000 mu. That is, it was found that the amount of cultivated land collected by the pilot censuses was closer to the data shown by current statistics, against the great difference in the evaluation by researchers and, in particular, with the estimate of the Department of Land Management.
The reasons for this are the following: (i) the measurement units used were sometimes different: In the land management survey, standard measurement units (mu) were used, whereas in the current statistics, and in the replies from the holders, the concept of traditional (productive) mu was often used (by which area is based on the quality of the land and its productive potential); (ii) the members of the household had different ideas (and replies) concerning the cultivated land areas, and the result depended on who the respondent was; (iii) certain agricultural holdings did not know the exact size of the land they were cultivating, because of the different systems of land use (contracted, transferred from parents, rented, etc.) and there was difficulty in remembering the number of parcels, and the size of each parcel; (iv) there was confusion between the right of ownership and the right of the use of land; (v) the frequent small land adjustments linked to the changes in family size; (vi) the brief terms of the transfer of the right to use land; (vii) the implementation of a new system of management organization for the use of land that was not operational in all counties; and (viii) holders were afraid of taxes.
Therefore, it was decided to estimate the agricultural land area using the results of the National Land Detailed Surveys (LDS)[3][3]. A new questionnaire, the “Card for Agricultural Land” (A606), was utilized. It was based on the original results from the up-dated LDS, provided by the National Land Administration. It obtained the agricultural land area by village[4][4] through the land validation survey carried out by the census.
The agricultural land data obtained from the questionnaires for household and non-household holdings (A601 and A602) were collected essentially to link this information with that for use of land, and on crop planting by each holding. This would allow estimating the distribution of land for the different uses, under the hypothesis that underestimation of the total land affects in the same proportion the information on the different uses of land.
The results of the census show that the data on cultivated land in China from questionnaires A601 and A602 are in thousands of hectares (86 317), which is close to the current statistics data of 95 466, and confirms the above-mentioned results of the pilot censuses. Therefore, the differences between the data from questionnaire A606 (130 039), and the data from questionnaires A601 and A602 cannot be ascribed only to the technical reasons mentioned in the atlas (different measurement units, calculation of ditches, etc. and actual treatment of steep slopes and stone rifts)[5][5]. After verification, the data from questionnaire A606 is considered more reliable than those reported by respondents.
For these reasons, there are doubts about the possible underestimation of the sown area (computed using data from forms A601 and A602). It is therefore important to carry out further analyses using the appropriate indicators and ratios to verify the consistency of this data with the other data collected.
3.4 Some Suggestions for the Next Agriculture Census and Current Statistics
The next agricultural census in China will certainly benefit from the crucial experience accumulated during the first census. As census takers have to deal with a quickly changing environment in China, the next census will be treated as a new exercise. In other words, the entire process will have to be opportunely and thoroughly tried and tested again.
Certain adjustments may be necessary, in terms of census scope, clear definitions and classification of the units according to the changes in the operational management of the holdings, and of the rules concerning contracted land (the number of the holdings is decreasing, moreover, holders try to abandon bad plots and to obtain better quality of land). In order to make timely decisions for the census design, it is opportune to implement and conduct the following:
1. 1. Continuously up-date the lists of the units created by the census, and prepare reliable detailed maps at the district and sub-district level, both as a frame for the current sample surveys, and to be used at the start of the next census.
2. 2. Verify changes in the organization and management of the agricultural and rural sector, and in the characteristics of agricultural production, to eventually conduct specific (sample) surveys.
3. 3. Carefully consider the possibility of conducting the census both by complete enumeration (with a short form questionnaire), and by sample survey (long form), as suggested by the expert consultation meeting held in 1994.
4. 4. Conduct a large experiment, with a sample survey specifically designed, to measure directly in the field the land area of holdings, and their utilization (as suggested during the International Seminar taken in 1995).
5. 5. Up-date the technical knowledge of the statistical staff, and implement specific training courses in the different fields.
6. 6. Up-date the available hardware and software, and follow the technological development in the different fields (also for data capture).
Moreover, certain issues could surface in organizing the next census, caused by improvements in the administrative system, organization and competence as a result of economic reform. The statistical infrastructure may run into difficulty in respondents’ cooperation, and finding the sufficient number of personnel.
Finally, considerations presented in this paragraph, and, in particular, the linkage between farm households and rural industries highlight the importance of developing: (i) integration (consistency) between agricultural current statistical surveys, and the Agricultural Census (Vogel F. A., 1999); (ii) the Agricultural Census in the context of a national integrated system of agricultural statistics must be enhanced. Chinese colleagues are aware of this and clearly have this objective (Zhu, 1998), and a paper on this topic will be presented at the seminar. However, it is recommended that they prepare a detailed multi-year plan for a sequential statistical programme of activities in order to verify the needs for statistical staff and infrastructure, and allocate adequate resources in a balanced manner to the components of the programme. This exercise would indicate the links among the different surveys, and the possibility of identifying the best data capture techniques and strategies.
4. Concluding remarks
Concluding, the analysis and research on the validity and quality of the census data, in comparison with current statistics, must be continued and enhanced with substantial user involvement, if necessary, to revise the published data. It is essential to ascertain that the data collected, and the figures disseminated, are reliable, and that users can access them easily.
Two innovative choices made by the FAO projects for the dissemination of the 1 percent sample database to all potential users on the Internet will fulfill this need: First, it was decided to set up an interactive analysis system, through the Statistical Data Analysis programmes (SDA), presenting a query facility through which tailored tabulations and statistical computations can be requested and performed immediately with no direct access to individual records, and no possibility of statistical disclosure. Second, the implemented GIS census data system for the production of census maps at the county level will operate in a distributed environment allowing users to obtain self-tailored maps. Obviously, this data set could be updated and integrated with other agricultural and economic data, as well as with information on infrastructure, etc. to improve its descriptive ability (NASS-USDA). These initiatives are useful for the economic data analyses of the census data but they will also foster analyses and research on the quality and validity of the data disseminated.
References
Biggeri M. (1998) The factors of the Chinese agricultural development after the reform: An empirical analysis, PhD thesis at the University of Siena, Italy (in Italian).
Cao Qingbo. (1999) China’s Agricultural Data and Statistical System and the 1997 Agricultural Census, (non-published paper).
FAO. (1986) Programme for the 1990 World Census of Agriculture, FAO Statistical Development Series 2, Rome.
FASC. (1995) Papers presented at the International Seminar on Taking the first agricultural census in China: Aims, problems, methodologies, challenges, opportunities, (Beijing, 13-17 November 1995).
NACOC & FASC. (1999) Abstract of the First National Agricultural Census in China, China Statistics Press, Beijing, China.
NASS-USDA, Data warehousing at the national agricultural statistics service. Easy & Fast Data Access for Everyone.
Onate B. T. (1995) Non-sampling errors (NSEs) in agricultural censuses and surveys: A major concern in the third world, Proceedings of 50th ISI Session, Beijing, China.
Rawski T. G. & Mead R. W. (1998) On the trail of China’s phantom farmers, World Development, vol. 26, n. 5, pp. 767-781, UK.
Razelle S. (1994) Decision-making in China’s rural economy: The linkage between villages leaders and farm household, The China Quarterly, n. 137, March 1994, pp. 99-124.
Shao Zongming, editor (1992) Proceedings of the international expert consultation on agricultural censuses, Beijing, China, Beijing Agricultural University Press.
SSB (1994), Programme for the nationwide pilot census of agricultural, China
Vogel F. A. (1999) The integration of agricultural sample survey and census methodologies, Bulletin of the ISI, Proceedings of the 52nd Session, Helsinki, Book 2, pp. 359-362.
Zhu Xiangdong, editor (1992) A selection of reports on the summing-up seminar of the Project GCP/CPR/006/ITA, Beijing Agricultural University Press.
Zhu Xiangdong. (1995) Preparation for 1997 census of agricultural in China, Proceedings of 50th ISI Session, Beijing, China.
Zhu Xiangdong. (1998) Agricultural census and improvement of rural statistics in China, paper presented at the Agricultural Statistics 2000, International Conference on Agricultural Statistics, March 18-20 1998, USA, organized by NASS, USDA.
[1][1] Use was made of project reports, analyses and work done by Chinese colleagues and consultants and coordinated by the author, who is solely responsible for the contents of this paper.
[2][2] An information system encompasses all the processes required to produce information relevant to the question in hand; by viewing data in a decision-making context, its acquired value became clear.
[3][3] The LDS was conducted according to common technical standards and rules: applied with air photos and large-scale topographic maps, and on-the-spot field survey.
[4][4] For the agricultural machinery and equipment, it was decided to collect the data at the village level, considering the difficulty in obtaining valid replies from the households in relation to the Chinese organization of the use of this equipment (as was confirmed by pilot census experiments).
[5][5] The differences are greater and of different proportions at the provincial level.