82. The organiser of this session was Mr. álvaro González-Villalobos, Senior Officer, FAO Statistics Division; and the discussant was Mr. I. P. David, Assistant Chief Economist, Asian Development Bank.

 

83. Three papers were presented:

(i)“Sample Rotation Using the Results of Agricultural Census” (Xu Lesheng);

(ii)“Survey Methodology for the Sample Census of Agriculture in Nigeria: Comparisons with  Experiences in Other Countries” (O. O. Ajayi);

(iii)“Improvement of Sample Design by Using Results from the Agricultural Census” (M. Steiner).

 

84. The first paper described the use of the census data to re-design the Rural Survey of China, incorporating sample rotation methods for first-stage sampling units.

 

85. The second paper dealt with the survey design adopted for the Census of Agriculture in Nigeria, in which the small-scale, subsistence farmers were covered on a sample basis while the large-scale farmers using modern implements were covered by complete enumeration. The data collection combined the objective measurement approach with face-to-face interviewing of farmers. The paper also compared this design with the methods used in PRC and some African countries. Suggestions for the FAO World Census of Agriculture Program were also made.

 

86. The third paper provided an overview of current Chinese agricultural data collection and sampling methods contrasted with a Pilot Sample Survey Program that was being developed. Under the current design, separate samples of rural households are often utilized to meet national, provincial, and local statistical data needs. It was reported that the Pilot Sample Survey utilized a replicated, multi-stage, multivariate probability proportional to size sample design. The design used the Census of Agriculture data to focus samples in villages where high concentrations of the items of interest existed at the time of the census. The utilization of multiple variables to create measures of size allowed specific crops, livestock and economic items to be targeted for sample selection.

 

87. Comments were made on the need to carry out a complete enumeration instead of a sample census. Many proposals were made on this matter. Other comments referred to data quality control procedures in the survey operations, the need for conducting further studies on variances to properly support the survey designs to be implemented, and the need to avoid overly complex sample designs. A number of questions addressed the importance of also considering Current Agricultural Surveys in PRC based on an area/multiple frame sample survey design. The three speakers agreed on the importance of conducting studies to eventually establish a Current Agricultural Survey Program in PRC based on such a design.