National Sample Census of Agriculture 1991-1992

CENSUS METHODOLOGY

A. Complete Enumeration of Holdings, Area and Livestock

The census methodology in undertaking the 1991/92 National Sample Census of Agriculture of Nepal is a combination of complete enumeration and sampling.

The Census of Agriculture was implemented in two phases. The first phase was the complete enumeration of all holdings, their area and the number of livestock and poultry raised by all households in the country as an integral activity in the listing operations of the Census of Population 1991. Two questions were asked concerning agricultural activities of households, namely: total area of the agriculture holding and total number of livestock/poultry kept by the households. All households listed in the Population Census 1991 that possessed the characteristics of the holding as defined in the agriculture census were identified as agricultural holdings. These holdings and their corresponding areas and number of livestock kept including their household population were compiled by ward for all the 75 districts of Nepal from which the sampling frame was constructed.

B. Sampling Design

A two-stage stratified sampling was employed in the selection of the samples for enumeration to obtain the characteristics of the holdings for the NSCA 1991. This design is similar to that of the 1981/82 sampling design, which is a self-weighting sample.

Construction of the sampling frame

The listing of the wards in each district with the summarized data of the number of holdings and area was used to form enumeration areas (EA's). However, wards containing less than 30 holdings were combined to form one EA. The EAs in each district were stratified according to the number of holdings enumerated, arranged from the highest to the lowest.

Selection of samples

The basic sampling methodology used was two stage area sampling as below:

First stage : selection of a stratified sample of enumeration areas (EAs) with probability proportional to expected number of holdings (stratified PPS sampling). i.e. the first stages of sample selection involved the primary sampling units (PSUs), where sample enumeration areas (EAs) were selected with probability proportional to size (PPS). The measure of size is the number of holdings enumerated in the EAs during the Census of Population 1991 listing operations.

To measure the importance of each district, the total area under 8 major crops was determined (paddy, wheat, maize, millet, barley, sugarcane, oilseed and potato). Districts were stratified into four groups according to this criterion. Group I represented the 10 least important districts; Group II, the next 15 important districts and Group III, the next 25 important districts and Group IV, the 25 most important districts. The number of selected EAs per district in each group follows:

    Group I - 50 EAs per district (total of 450 EAs because one district, specifically Manang district was taken as a certainty sample district)
    Group II - 60 EAs per district (total of 900 EAs)
    Group III - 70 EAs per district (total of 1,750 EAs)
    Group IV - 80 EAs per district (total of 2000 EAs).

Second stage : within selected EAs, the selection of a sample of agricultural holdings using stratified systematic random sampling. That is, the second stage of sample selection involved the selection of sample holdings systematically in each sample EA. Before the sample selection was done, a listing of holdings in each sample PSU was conducted to update the listing during the Population Census. The target number of holdings for enumeration in each sample EA was 25.

Altogether, 5,100 enumeration areas were selected from the 74 districts. The final sample of holdings for the country was about 122,000. A complete enumeration of all holdings was carried out in Manang District.

The details of the sampling procedures and the estimation of parameters for each district are found in the RELIABILITY OF DATA, which is one of the series of reports prepared for reliability of NSCA 1991/92.